Use 32-bit big write location (OK for up to 512 TB OSDs)

This is required to:
1) later inline the last "big_write" entry into the object to slightly reduce memory usage
2) eliminate an ugly hack where entry type is determined by its size
3) make the storage scheme extensible i.e. when adding new entry types

.gitea/workflows/buildenv.Dockerfile

@@ -1,29 +1,28 @@
-FROM node:16-bookworm
+FROM node:16-bullseye
 
 WORKDIR /root
 
-ADD ./docker/etc/apt/trusted.gpg.d /etc/apt/trusted.gpg.d
+ADD ./docker/vitastor.gpg /etc/apt/trusted.gpg.d
 
-RUN echo 'deb http://deb.debian.org/debian bookworm-backports main' >> /etc/apt/sources.list; \
-    echo 'deb http://vitastor.io/debian bookworm main' >> /etc/apt/sources.list; \
+RUN echo 'deb http://deb.debian.org/debian bullseye-backports main' >> /etc/apt/sources.list; \
+    echo 'deb http://vitastor.io/debian bullseye main' >> /etc/apt/sources.list; \
     echo >> /etc/apt/preferences; \
     echo 'Package: *' >> /etc/apt/preferences; \
-    echo 'Pin: release n=bookworm-backports' >> /etc/apt/preferences; \
+    echo 'Pin: release a=bullseye-backports' >> /etc/apt/preferences; \
     echo 'Pin-Priority: 500' >> /etc/apt/preferences; \
     echo >> /etc/apt/preferences; \
     echo 'Package: *' >> /etc/apt/preferences; \
     echo 'Pin: origin "vitastor.io"' >> /etc/apt/preferences; \
     echo 'Pin-Priority: 1000' >> /etc/apt/preferences; \
-    perl -i -pe 's/Types: deb$/Types: deb deb-src/' /etc/apt/sources.list.d/debian.sources; \
     grep '^deb ' /etc/apt/sources.list | perl -pe 's/^deb/deb-src/' >> /etc/apt/sources.list; \
     echo 'APT::Install-Recommends false;' >> /etc/apt/apt.conf; \
     echo 'APT::Install-Suggests false;' >> /etc/apt/apt.conf
 
 RUN apt-get update
-RUN apt-get -y install etcd qemu-system-x86 qemu-block-extra qemu-utils fio libasan8 \
+RUN apt-get -y install etcd qemu-system-x86 qemu-block-extra qemu-utils fio libasan5 \
     libgoogle-perftools-dev devscripts libjerasure-dev cmake libibverbs-dev libisal-dev
 RUN apt-get -y build-dep fio qemu=`dpkg -s qemu-system-x86|grep ^Version:|awk '{print $2}'`
-RUN apt-get update && apt-get -y install jq lp-solve sudo nfs-common fdisk parted libc-ares-dev udev
+RUN apt-get update && apt-get -y install jq lp-solve sudo nfs-common fdisk parted
 RUN apt-get --download-only source fio qemu=`dpkg -s qemu-system-x86|grep ^Version:|awk '{print $2}'`
 
 RUN set -ex; \

.gitea/workflows/test.yml

@@ -63,7 +63,7 @@ jobs:
     container: ${{env.TEST_IMAGE}}:${{github.sha}}
     steps:
     # leak sanitizer sometimes crashes
-    - run: cd /root/vitastor/build && ASAN_OPTIONS=detect_leaks=0 make -j16 build_tests test
+    - run: cd /root/vitastor/build && ASAN_OPTIONS=detect_leaks=0 make -j16 test
 
   npm_lint:
     runs-on: ubuntu-latest
@@ -306,78 +306,6 @@ jobs:
           echo ""
         done
 
-  test_old_interrupted_rebalance:
-    runs-on: ubuntu-latest
-    needs: build
-    container: ${{env.TEST_IMAGE}}:${{github.sha}}
-    steps:
-    - name: Run test
-      id: test
-      timeout-minutes: 10
-      run: OLD=1 /root/vitastor/tests/test_interrupted_rebalance.sh
-    - name: Print logs
-      if: always() && steps.test.outcome == 'failure'
-      run: |
-        for i in /root/vitastor/testdata/*.log /root/vitastor/testdata/*.txt; do
-          echo "-------- $i --------"
-          cat $i
-          echo ""
-        done
-
-  test_old_interrupted_rebalance_imm:
-    runs-on: ubuntu-latest
-    needs: build
-    container: ${{env.TEST_IMAGE}}:${{github.sha}}
-    steps:
-    - name: Run test
-      id: test
-      timeout-minutes: 10
-      run: OLD=1 IMMEDIATE_COMMIT=1 /root/vitastor/tests/test_interrupted_rebalance.sh
-    - name: Print logs
-      if: always() && steps.test.outcome == 'failure'
-      run: |
-        for i in /root/vitastor/testdata/*.log /root/vitastor/testdata/*.txt; do
-          echo "-------- $i --------"
-          cat $i
-          echo ""
-        done
-
-  test_old_interrupted_rebalance_ec:
-    runs-on: ubuntu-latest
-    needs: build
-    container: ${{env.TEST_IMAGE}}:${{github.sha}}
-    steps:
-    - name: Run test
-      id: test
-      timeout-minutes: 10
-      run: OLD=1 SCHEME=ec /root/vitastor/tests/test_interrupted_rebalance.sh
-    - name: Print logs
-      if: always() && steps.test.outcome == 'failure'
-      run: |
-        for i in /root/vitastor/testdata/*.log /root/vitastor/testdata/*.txt; do
-          echo "-------- $i --------"
-          cat $i
-          echo ""
-        done
-
-  test_old_interrupted_rebalance_ec_imm:
-    runs-on: ubuntu-latest
-    needs: build
-    container: ${{env.TEST_IMAGE}}:${{github.sha}}
-    steps:
-    - name: Run test
-      id: test
-      timeout-minutes: 10
-      run: OLD=1 SCHEME=ec IMMEDIATE_COMMIT=1 /root/vitastor/tests/test_interrupted_rebalance.sh
-    - name: Print logs
-      if: always() && steps.test.outcome == 'failure'
-      run: |
-        for i in /root/vitastor/testdata/*.log /root/vitastor/testdata/*.txt; do
-          echo "-------- $i --------"
-          cat $i
-          echo ""
-        done
-
   test_create_halfhost:
     runs-on: ubuntu-latest
     needs: build
@@ -414,24 +342,6 @@ jobs:
           echo ""
         done
 
-  test_level_placement:
-    runs-on: ubuntu-latest
-    needs: build
-    container: ${{env.TEST_IMAGE}}:${{github.sha}}
-    steps:
-    - name: Run test
-      id: test
-      timeout-minutes: 3
-      run: /root/vitastor/tests/test_level_placement.sh
-    - name: Print logs
-      if: always() && steps.test.outcome == 'failure'
-      run: |
-        for i in /root/vitastor/testdata/*.log /root/vitastor/testdata/*.txt; do
-          echo "-------- $i --------"
-          cat $i
-          echo ""
-        done
-
   test_snapshot:
     runs-on: ubuntu-latest
     needs: build
@@ -468,42 +378,6 @@ jobs:
           echo ""
         done
 
-  test_old_snapshot:
-    runs-on: ubuntu-latest
-    needs: build
-    container: ${{env.TEST_IMAGE}}:${{github.sha}}
-    steps:
-    - name: Run test
-      id: test
-      timeout-minutes: 3
-      run: OLD=1 /root/vitastor/tests/test_snapshot.sh
-    - name: Print logs
-      if: always() && steps.test.outcome == 'failure'
-      run: |
-        for i in /root/vitastor/testdata/*.log /root/vitastor/testdata/*.txt; do
-          echo "-------- $i --------"
-          cat $i
-          echo ""
-        done
-
-  test_old_snapshot_ec:
-    runs-on: ubuntu-latest
-    needs: build
-    container: ${{env.TEST_IMAGE}}:${{github.sha}}
-    steps:
-    - name: Run test
-      id: test
-      timeout-minutes: 3
-      run: OLD=1 SCHEME=ec /root/vitastor/tests/test_snapshot.sh
-    - name: Print logs
-      if: always() && steps.test.outcome == 'failure'
-      run: |
-        for i in /root/vitastor/testdata/*.log /root/vitastor/testdata/*.txt; do
-          echo "-------- $i --------"
-          cat $i
-          echo ""
-        done
-
   test_minsize_1:
     runs-on: ubuntu-latest
     needs: build
@@ -540,24 +414,6 @@ jobs:
           echo ""
         done
 
-  test_old_move_reappear:
-    runs-on: ubuntu-latest
-    needs: build
-    container: ${{env.TEST_IMAGE}}:${{github.sha}}
-    steps:
-    - name: Run test
-      id: test
-      timeout-minutes: 3
-      run: OLD=1 /root/vitastor/tests/test_move_reappear.sh
-    - name: Print logs
-      if: always() && steps.test.outcome == 'failure'
-      run: |
-        for i in /root/vitastor/testdata/*.log /root/vitastor/testdata/*.txt; do
-          echo "-------- $i --------"
-          cat $i
-          echo ""
-        done
-
   test_degraded:
     runs-on: ubuntu-latest
     needs: build
@@ -648,42 +504,6 @@ jobs:
           echo ""
         done
 
-  test_old_snapshot_chain:
-    runs-on: ubuntu-latest
-    needs: build
-    container: ${{env.TEST_IMAGE}}:${{github.sha}}
-    steps:
-    - name: Run test
-      id: test
-      timeout-minutes: 3
-      run: OLD=1 /root/vitastor/tests/test_snapshot_chain.sh
-    - name: Print logs
-      if: always() && steps.test.outcome == 'failure'
-      run: |
-        for i in /root/vitastor/testdata/*.log /root/vitastor/testdata/*.txt; do
-          echo "-------- $i --------"
-          cat $i
-          echo ""
-        done
-
-  test_old_snapshot_chain_ec:
-    runs-on: ubuntu-latest
-    needs: build
-    container: ${{env.TEST_IMAGE}}:${{github.sha}}
-    steps:
-    - name: Run test
-      id: test
-      timeout-minutes: 3
-      run: OLD=1 SCHEME=ec /root/vitastor/tests/test_snapshot_chain.sh
-    - name: Print logs
-      if: always() && steps.test.outcome == 'failure'
-      run: |
-        for i in /root/vitastor/testdata/*.log /root/vitastor/testdata/*.txt; do
-          echo "-------- $i --------"
-          cat $i
-          echo ""
-        done
-
   test_snapshot_down:
     runs-on: ubuntu-latest
     needs: build
@@ -720,78 +540,6 @@ jobs:
           echo ""
         done
 
-  test_old_snapshot_down:
-    runs-on: ubuntu-latest
-    needs: build
-    container: ${{env.TEST_IMAGE}}:${{github.sha}}
-    steps:
-    - name: Run test
-      id: test
-      timeout-minutes: 3
-      run: OLD=1 /root/vitastor/tests/test_snapshot_down.sh
-    - name: Print logs
-      if: always() && steps.test.outcome == 'failure'
-      run: |
-        for i in /root/vitastor/testdata/*.log /root/vitastor/testdata/*.txt; do
-          echo "-------- $i --------"
-          cat $i
-          echo ""
-        done
-
-  test_old_snapshot_down_ec:
-    runs-on: ubuntu-latest
-    needs: build
-    container: ${{env.TEST_IMAGE}}:${{github.sha}}
-    steps:
-    - name: Run test
-      id: test
-      timeout-minutes: 3
-      run: OLD=1 SCHEME=ec /root/vitastor/tests/test_snapshot_down.sh
-    - name: Print logs
-      if: always() && steps.test.outcome == 'failure'
-      run: |
-        for i in /root/vitastor/testdata/*.log /root/vitastor/testdata/*.txt; do
-          echo "-------- $i --------"
-          cat $i
-          echo ""
-        done
-
-  test_kv_stress:
-    runs-on: ubuntu-latest
-    needs: build
-    container: ${{env.TEST_IMAGE}}:${{github.sha}}
-    steps:
-    - name: Run test
-      id: test
-      timeout-minutes: 3
-      run: /root/vitastor/tests/test_kv_stress.sh
-    - name: Print logs
-      if: always() && steps.test.outcome == 'failure'
-      run: |
-        for i in /root/vitastor/testdata/*.log /root/vitastor/testdata/*.txt; do
-          echo "-------- $i --------"
-          cat $i
-          echo ""
-        done
-
-  test_kv_stress_imm:
-    runs-on: ubuntu-latest
-    needs: build
-    container: ${{env.TEST_IMAGE}}:${{github.sha}}
-    steps:
-    - name: Run test
-      id: test
-      timeout-minutes: 3
-      run: IMMEDIATE_COMMIT=1 /root/vitastor/tests/test_kv_stress.sh
-    - name: Print logs
-      if: always() && steps.test.outcome == 'failure'
-      run: |
-        for i in /root/vitastor/testdata/*.log /root/vitastor/testdata/*.txt; do
-          echo "-------- $i --------"
-          cat $i
-          echo ""
-        done
-
   test_splitbrain:
     runs-on: ubuntu-latest
     needs: build
@@ -882,78 +630,6 @@ jobs:
           echo ""
         done
 
-  test_old_rebalance_verify:
-    runs-on: ubuntu-latest
-    needs: build
-    container: ${{env.TEST_IMAGE}}:${{github.sha}}
-    steps:
-    - name: Run test
-      id: test
-      timeout-minutes: 10
-      run: OLD=1 /root/vitastor/tests/test_rebalance_verify.sh
-    - name: Print logs
-      if: always() && steps.test.outcome == 'failure'
-      run: |
-        for i in /root/vitastor/testdata/*.log /root/vitastor/testdata/*.txt; do
-          echo "-------- $i --------"
-          cat $i
-          echo ""
-        done
-
-  test_old_rebalance_verify_imm:
-    runs-on: ubuntu-latest
-    needs: build
-    container: ${{env.TEST_IMAGE}}:${{github.sha}}
-    steps:
-    - name: Run test
-      id: test
-      timeout-minutes: 10
-      run: OLD=1 IMMEDIATE_COMMIT=1 /root/vitastor/tests/test_rebalance_verify.sh
-    - name: Print logs
-      if: always() && steps.test.outcome == 'failure'
-      run: |
-        for i in /root/vitastor/testdata/*.log /root/vitastor/testdata/*.txt; do
-          echo "-------- $i --------"
-          cat $i
-          echo ""
-        done
-
-  test_old_rebalance_verify_ec:
-    runs-on: ubuntu-latest
-    needs: build
-    container: ${{env.TEST_IMAGE}}:${{github.sha}}
-    steps:
-    - name: Run test
-      id: test
-      timeout-minutes: 10
-      run: OLD=1 SCHEME=ec /root/vitastor/tests/test_rebalance_verify.sh
-    - name: Print logs
-      if: always() && steps.test.outcome == 'failure'
-      run: |
-        for i in /root/vitastor/testdata/*.log /root/vitastor/testdata/*.txt; do
-          echo "-------- $i --------"
-          cat $i
-          echo ""
-        done
-
-  test_old_rebalance_verify_ec_imm:
-    runs-on: ubuntu-latest
-    needs: build
-    container: ${{env.TEST_IMAGE}}:${{github.sha}}
-    steps:
-    - name: Run test
-      id: test
-      timeout-minutes: 10
-      run: OLD=1 SCHEME=ec IMMEDIATE_COMMIT=1 /root/vitastor/tests/test_rebalance_verify.sh
-    - name: Print logs
-      if: always() && steps.test.outcome == 'failure'
-      run: |
-        for i in /root/vitastor/testdata/*.log /root/vitastor/testdata/*.txt; do
-          echo "-------- $i --------"
-          cat $i
-          echo ""
-        done
-
   test_dd:
     runs-on: ubuntu-latest
     needs: build
@@ -1062,7 +738,7 @@ jobs:
           echo ""
         done
 
-  test_old_write:
+  test_write_no_same:
     runs-on: ubuntu-latest
     needs: build
     container: ${{env.TEST_IMAGE}}:${{github.sha}}
@@ -1070,61 +746,7 @@ jobs:
     - name: Run test
       id: test
       timeout-minutes: 3
-      run: OLD=1 /root/vitastor/tests/test_write.sh
-    - name: Print logs
-      if: always() && steps.test.outcome == 'failure'
-      run: |
-        for i in /root/vitastor/testdata/*.log /root/vitastor/testdata/*.txt; do
-          echo "-------- $i --------"
-          cat $i
-          echo ""
-        done
-
-  test_old_write_xor:
-    runs-on: ubuntu-latest
-    needs: build
-    container: ${{env.TEST_IMAGE}}:${{github.sha}}
-    steps:
-    - name: Run test
-      id: test
-      timeout-minutes: 3
-      run: OLD=1 SCHEME=xor /root/vitastor/tests/test_write.sh
-    - name: Print logs
-      if: always() && steps.test.outcome == 'failure'
-      run: |
-        for i in /root/vitastor/testdata/*.log /root/vitastor/testdata/*.txt; do
-          echo "-------- $i --------"
-          cat $i
-          echo ""
-        done
-
-  test_write_old_iothreads:
-    runs-on: ubuntu-latest
-    needs: build
-    container: ${{env.TEST_IMAGE}}:${{github.sha}}
-    steps:
-    - name: Run test
-      id: test
-      timeout-minutes: 3
-      run: TEST_NAME=old_iothreads OLD=1 GLOBAL_CONFIG=',"client_iothread_count":4' /root/vitastor/tests/test_write.sh
-    - name: Print logs
-      if: always() && steps.test.outcome == 'failure'
-      run: |
-        for i in /root/vitastor/testdata/*.log /root/vitastor/testdata/*.txt; do
-          echo "-------- $i --------"
-          cat $i
-          echo ""
-        done
-
-  test_old_write_no_same:
-    runs-on: ubuntu-latest
-    needs: build
-    container: ${{env.TEST_IMAGE}}:${{github.sha}}
-    steps:
-    - name: Run test
-      id: test
-      timeout-minutes: 3
-      run: OLD=1 /root/vitastor/tests/test_write_no_same.sh
+      run: /root/vitastor/tests/test_write_no_same.sh
     - name: Print logs
       if: always() && steps.test.outcome == 'failure'
       run: |
@@ -1206,78 +828,6 @@ jobs:
           echo ""
         done
 
-  test_checksum:
-    runs-on: ubuntu-latest
-    needs: build
-    container: ${{env.TEST_IMAGE}}:${{github.sha}}
-    steps:
-    - name: Run test
-      id: test
-      timeout-minutes: 3
-      run: /root/vitastor/tests/test_checksum.sh
-    - name: Print logs
-      if: always() && steps.test.outcome == 'failure'
-      run: |
-        for i in /root/vitastor/testdata/*.log /root/vitastor/testdata/*.txt; do
-          echo "-------- $i --------"
-          cat $i
-          echo ""
-        done
-
-  test_old_checksum:
-    runs-on: ubuntu-latest
-    needs: build
-    container: ${{env.TEST_IMAGE}}:${{github.sha}}
-    steps:
-    - name: Run test
-      id: test
-      timeout-minutes: 3
-      run: OLD=1 /root/vitastor/tests/test_checksum.sh
-    - name: Print logs
-      if: always() && steps.test.outcome == 'failure'
-      run: |
-        for i in /root/vitastor/testdata/*.log /root/vitastor/testdata/*.txt; do
-          echo "-------- $i --------"
-          cat $i
-          echo ""
-        done
-
-  test_corrupt_all:
-    runs-on: ubuntu-latest
-    needs: build
-    container: ${{env.TEST_IMAGE}}:${{github.sha}}
-    steps:
-    - name: Run test
-      id: test
-      timeout-minutes: 3
-      run: /root/vitastor/tests/test_corrupt_all.sh
-    - name: Print logs
-      if: always() && steps.test.outcome == 'failure'
-      run: |
-        for i in /root/vitastor/testdata/*.log /root/vitastor/testdata/*.txt; do
-          echo "-------- $i --------"
-          cat $i
-          echo ""
-        done
-
-  test_old_corrupt_all:
-    runs-on: ubuntu-latest
-    needs: build
-    container: ${{env.TEST_IMAGE}}:${{github.sha}}
-    steps:
-    - name: Run test
-      id: test
-      timeout-minutes: 3
-      run: OLD=1 /root/vitastor/tests/test_corrupt_all.sh
-    - name: Print logs
-      if: always() && steps.test.outcome == 'failure'
-      run: |
-        for i in /root/vitastor/testdata/*.log /root/vitastor/testdata/*.txt; do
-          echo "-------- $i --------"
-          cat $i
-          echo ""
-        done
-
   test_reweight_half:
     runs-on: ubuntu-latest
     needs: build
@@ -1476,42 +1026,6 @@ jobs:
           echo ""
         done
 
-  test_old_resize:
-    runs-on: ubuntu-latest
-    needs: build
-    container: ${{env.TEST_IMAGE}}:${{github.sha}}
-    steps:
-    - name: Run test
-      id: test
-      timeout-minutes: 3
-      run: OLD=1 /root/vitastor/tests/test_resize.sh
-    - name: Print logs
-      if: always() && steps.test.outcome == 'failure'
-      run: |
-        for i in /root/vitastor/testdata/*.log /root/vitastor/testdata/*.txt; do
-          echo "-------- $i --------"
-          cat $i
-          echo ""
-        done
-
-  test_old_resize_auto:
-    runs-on: ubuntu-latest
-    needs: build
-    container: ${{env.TEST_IMAGE}}:${{github.sha}}
-    steps:
-    - name: Run test
-      id: test
-      timeout-minutes: 3
-      run: OLD=1 /root/vitastor/tests/test_resize_auto.sh
-    - name: Print logs
-      if: always() && steps.test.outcome == 'failure'
-      run: |
-        for i in /root/vitastor/testdata/*.log /root/vitastor/testdata/*.txt; do
-          echo "-------- $i --------"
-          cat $i
-          echo ""
-        done
-
   test_osd_tags:
     runs-on: ubuntu-latest
     needs: build
@@ -1602,78 +1116,6 @@ jobs:
           echo ""
         done
 
-  test_old_enospc:
-    runs-on: ubuntu-latest
-    needs: build
-    container: ${{env.TEST_IMAGE}}:${{github.sha}}
-    steps:
-    - name: Run test
-      id: test
-      timeout-minutes: 3
-      run: OLD=1 /root/vitastor/tests/test_enospc.sh
-    - name: Print logs
-      if: always() && steps.test.outcome == 'failure'
-      run: |
-        for i in /root/vitastor/testdata/*.log /root/vitastor/testdata/*.txt; do
-          echo "-------- $i --------"
-          cat $i
-          echo ""
-        done
-
-  test_old_enospc_xor:
-    runs-on: ubuntu-latest
-    needs: build
-    container: ${{env.TEST_IMAGE}}:${{github.sha}}
-    steps:
-    - name: Run test
-      id: test
-      timeout-minutes: 3
-      run: OLD=1 SCHEME=xor /root/vitastor/tests/test_enospc.sh
-    - name: Print logs
-      if: always() && steps.test.outcome == 'failure'
-      run: |
-        for i in /root/vitastor/testdata/*.log /root/vitastor/testdata/*.txt; do
-          echo "-------- $i --------"
-          cat $i
-          echo ""
-        done
-
-  test_old_enospc_imm:
-    runs-on: ubuntu-latest
-    needs: build
-    container: ${{env.TEST_IMAGE}}:${{github.sha}}
-    steps:
-    - name: Run test
-      id: test
-      timeout-minutes: 3
-      run: OLD=1 IMMEDIATE_COMMIT=1 /root/vitastor/tests/test_enospc.sh
-    - name: Print logs
-      if: always() && steps.test.outcome == 'failure'
-      run: |
-        for i in /root/vitastor/testdata/*.log /root/vitastor/testdata/*.txt; do
-          echo "-------- $i --------"
-          cat $i
-          echo ""
-        done
-
-  test_old_enospc_imm_xor:
-    runs-on: ubuntu-latest
-    needs: build
-    container: ${{env.TEST_IMAGE}}:${{github.sha}}
-    steps:
-    - name: Run test
-      id: test
-      timeout-minutes: 3
-      run: OLD=1 IMMEDIATE_COMMIT=1 SCHEME=xor /root/vitastor/tests/test_enospc.sh
-    - name: Print logs
-      if: always() && steps.test.outcome == 'failure'
-      run: |
-        for i in /root/vitastor/testdata/*.log /root/vitastor/testdata/*.txt; do
-          echo "-------- $i --------"
-          cat $i
-          echo ""
-        done
-
   test_scrub:
     runs-on: ubuntu-latest
     needs: build
@@ -1782,240 +1224,6 @@ jobs:
           echo ""
         done
 
-  test_old_scrub:
-    runs-on: ubuntu-latest
-    needs: build
-    container: ${{env.TEST_IMAGE}}:${{github.sha}}
-    steps:
-    - name: Run test
-      id: test
-      timeout-minutes: 3
-      run: OLD=1 /root/vitastor/tests/test_scrub.sh
-    - name: Print logs
-      if: always() && steps.test.outcome == 'failure'
-      run: |
-        for i in /root/vitastor/testdata/*.log /root/vitastor/testdata/*.txt; do
-          echo "-------- $i --------"
-          cat $i
-          echo ""
-        done
-
-  test_old_scrub_zero_osd_2:
-    runs-on: ubuntu-latest
-    needs: build
-    container: ${{env.TEST_IMAGE}}:${{github.sha}}
-    steps:
-    - name: Run test
-      id: test
-      timeout-minutes: 3
-      run: OLD=1 ZERO_OSD=2 /root/vitastor/tests/test_scrub.sh
-    - name: Print logs
-      if: always() && steps.test.outcome == 'failure'
-      run: |
-        for i in /root/vitastor/testdata/*.log /root/vitastor/testdata/*.txt; do
-          echo "-------- $i --------"
-          cat $i
-          echo ""
-        done
-
-  test_old_scrub_xor:
-    runs-on: ubuntu-latest
-    needs: build
-    container: ${{env.TEST_IMAGE}}:${{github.sha}}
-    steps:
-    - name: Run test
-      id: test
-      timeout-minutes: 3
-      run: OLD=1 SCHEME=xor /root/vitastor/tests/test_scrub.sh
-    - name: Print logs
-      if: always() && steps.test.outcome == 'failure'
-      run: |
-        for i in /root/vitastor/testdata/*.log /root/vitastor/testdata/*.txt; do
-          echo "-------- $i --------"
-          cat $i
-          echo ""
-        done
-
-  test_old_scrub_pg_size_3:
-    runs-on: ubuntu-latest
-    needs: build
-    container: ${{env.TEST_IMAGE}}:${{github.sha}}
-    steps:
-    - name: Run test
-      id: test
-      timeout-minutes: 3
-      run: OLD=1 PG_SIZE=3 /root/vitastor/tests/test_scrub.sh
-    - name: Print logs
-      if: always() && steps.test.outcome == 'failure'
-      run: |
-        for i in /root/vitastor/testdata/*.log /root/vitastor/testdata/*.txt; do
-          echo "-------- $i --------"
-          cat $i
-          echo ""
-        done
-
-  test_old_scrub_pg_size_6_pg_minsize_4_osd_count_6_ec:
-    runs-on: ubuntu-latest
-    needs: build
-    container: ${{env.TEST_IMAGE}}:${{github.sha}}
-    steps:
-    - name: Run test
-      id: test
-      timeout-minutes: 3
-      run: OLD=1 PG_SIZE=6 PG_MINSIZE=4 OSD_COUNT=6 SCHEME=ec /root/vitastor/tests/test_scrub.sh
-    - name: Print logs
-      if: always() && steps.test.outcome == 'failure'
-      run: |
-        for i in /root/vitastor/testdata/*.log /root/vitastor/testdata/*.txt; do
-          echo "-------- $i --------"
-          cat $i
-          echo ""
-        done
-
-  test_old_scrub_ec:
-    runs-on: ubuntu-latest
-    needs: build
-    container: ${{env.TEST_IMAGE}}:${{github.sha}}
-    steps:
-    - name: Run test
-      id: test
-      timeout-minutes: 3
-      run: OLD=1 SCHEME=ec /root/vitastor/tests/test_scrub.sh
-    - name: Print logs
-      if: always() && steps.test.outcome == 'failure'
-      run: |
-        for i in /root/vitastor/testdata/*.log /root/vitastor/testdata/*.txt; do
-          echo "-------- $i --------"
-          cat $i
-          echo ""
-        done
-
-  test_old_partwr_csum:
-    runs-on: ubuntu-latest
-    needs: build
-    container: ${{env.TEST_IMAGE}}:${{github.sha}}
-    steps:
-    - name: Run test
-      id: test
-      timeout-minutes: 3
-      run: OLD=1 /root/vitastor/tests/test_partwr_csum.sh
-    - name: Print logs
-      if: always() && steps.test.outcome == 'failure'
-      run: |
-        for i in /root/vitastor/testdata/*.log /root/vitastor/testdata/*.txt; do
-          echo "-------- $i --------"
-          cat $i
-          echo ""
-        done
-
-  test_heal_old_csum_32k_dmj:
-    runs-on: ubuntu-latest
-    needs: build
-    container: ${{env.TEST_IMAGE}}:${{github.sha}}
-    steps:
-    - name: Run test
-      id: test
-      timeout-minutes: 10
-      run: TEST_NAME=old_csum_32k_dmj OLD=1 OSD_ARGS="--data_csum_type crc32c --csum_block_size 32k --inmemory_metadata false --inmemory_journal false" OFFSET_ARGS=$OSD_ARGS /root/vitastor/tests/test_heal.sh
-    - name: Print logs
-      if: always() && steps.test.outcome == 'failure'
-      run: |
-        for i in /root/vitastor/testdata/*.log /root/vitastor/testdata/*.txt; do
-          echo "-------- $i --------"
-          cat $i
-          echo ""
-        done
-
-  test_heal_old_csum_32k_dj:
-    runs-on: ubuntu-latest
-    needs: build
-    container: ${{env.TEST_IMAGE}}:${{github.sha}}
-    steps:
-    - name: Run test
-      id: test
-      timeout-minutes: 10
-      run: TEST_NAME=old_csum_32k_dj  OLD=1 OSD_ARGS="--data_csum_type crc32c --csum_block_size 32k --inmemory_journal false" OFFSET_ARGS=$OSD_ARGS /root/vitastor/tests/test_heal.sh
-    - name: Print logs
-      if: always() && steps.test.outcome == 'failure'
-      run: |
-        for i in /root/vitastor/testdata/*.log /root/vitastor/testdata/*.txt; do
-          echo "-------- $i --------"
-          cat $i
-          echo ""
-        done
-
-  test_heal_old_csum_32k:
-    runs-on: ubuntu-latest
-    needs: build
-    container: ${{env.TEST_IMAGE}}:${{github.sha}}
-    steps:
-    - name: Run test
-      id: test
-      timeout-minutes: 10
-      run: TEST_NAME=old_csum_32k     OLD=1 OSD_ARGS="--data_csum_type crc32c --csum_block_size 32k" OFFSET_ARGS=$OSD_ARGS /root/vitastor/tests/test_heal.sh
-    - name: Print logs
-      if: always() && steps.test.outcome == 'failure'
-      run: |
-        for i in /root/vitastor/testdata/*.log /root/vitastor/testdata/*.txt; do
-          echo "-------- $i --------"
-          cat $i
-          echo ""
-        done
-
-  test_heal_old_csum_4k_dmj:
-    runs-on: ubuntu-latest
-    needs: build
-    container: ${{env.TEST_IMAGE}}:${{github.sha}}
-    steps:
-    - name: Run test
-      id: test
-      timeout-minutes: 10
-      run: TEST_NAME=old_csum_4k_dmj  OLD=1 OSD_ARGS="--data_csum_type crc32c --inmemory_metadata false --inmemory_journal false" OFFSET_ARGS=$OSD_ARGS /root/vitastor/tests/test_heal.sh
-    - name: Print logs
-      if: always() && steps.test.outcome == 'failure'
-      run: |
-        for i in /root/vitastor/testdata/*.log /root/vitastor/testdata/*.txt; do
-          echo "-------- $i --------"
-          cat $i
-          echo ""
-        done
-
-  test_heal_old_csum_4k_dj:
-    runs-on: ubuntu-latest
-    needs: build
-    container: ${{env.TEST_IMAGE}}:${{github.sha}}
-    steps:
-    - name: Run test
-      id: test
-      timeout-minutes: 10
-      run: TEST_NAME=old_csum_4k_dj   OLD=1 OSD_ARGS="--data_csum_type crc32c --inmemory_journal false" OFFSET_ARGS=$OSD_ARGS /root/vitastor/tests/test_heal.sh
-    - name: Print logs
-      if: always() && steps.test.outcome == 'failure'
-      run: |
-        for i in /root/vitastor/testdata/*.log /root/vitastor/testdata/*.txt; do
-          echo "-------- $i --------"
-          cat $i
-          echo ""
-        done
-
-  test_heal_old_csum_4k:
-    runs-on: ubuntu-latest
-    needs: build
-    container: ${{env.TEST_IMAGE}}:${{github.sha}}
-    steps:
-    - name: Run test
-      id: test
-      timeout-minutes: 10
-      run: TEST_NAME=old_csum_4k      OLD=1 OSD_ARGS="--data_csum_type crc32c" OFFSET_ARGS=$OSD_ARGS /root/vitastor/tests/test_heal.sh
-    - name: Print logs
-      if: always() && steps.test.outcome == 'failure'
-      run: |
-        for i in /root/vitastor/testdata/*.log /root/vitastor/testdata/*.txt; do
-          echo "-------- $i --------"
-          cat $i
-          echo ""
-        done
-
   test_nfs:
     runs-on: ubuntu-latest
     needs: build
@@ -2034,21 +1242,3 @@ jobs:
           echo ""
         done
 
-  test_nfs_unaligned_append:
-    runs-on: ubuntu-latest
-    needs: build
-    container: ${{env.TEST_IMAGE}}:${{github.sha}}
-    steps:
-    - name: Run test
-      id: test
-      timeout-minutes: 3
-      run: /root/vitastor/tests/test_nfs_unaligned_append.sh
-    - name: Print logs
-      if: always() && steps.test.outcome == 'failure'
-      run: |
-        for i in /root/vitastor/testdata/*.log /root/vitastor/testdata/*.txt; do
-          echo "-------- $i --------"
-          cat $i
-          echo ""
-        done
-

.gitea/workflows/tests-to-yaml.pl

@@ -38,10 +38,6 @@ for my $line (<>)
             {
                 $test_name .= '_antietcd';
             }
-            elsif ($1 eq 'OLD')
-            {
-                $test_name =~ s/^test_/test_old_/s;
-            }
             else
             {
                 $test_name .= '_'.lc($1).'_'.$2;

CMakeLists.txt

@@ -1,20 +1,20 @@
-cmake_minimum_required(VERSION 2.8...3.30)
+cmake_minimum_required(VERSION 2.8.12)
 
 project(vitastor)
 
-set(VITASTOR_VERSION "3.0.12")
+set(VITASTOR_VERSION "2.4.3")
 
 include(CTest)
 
 add_custom_target(build_tests)
-set_property(TEST PROPERTY ENVIRONMENT LSAN_OPTIONS=suppressions=${CMAKE_CURRENT_BINARY_DIR}/lsan-suppress.txt)
-add_test(gen_lsan_suppress
-    ${CMAKE_COMMAND} -E echo leak:tcmalloc > "${CMAKE_CURRENT_BINARY_DIR}/lsan-suppress.txt"
+add_custom_target(test
+	COMMAND
+	echo leak:tcmalloc > ${CMAKE_CURRENT_BINARY_DIR}/lsan-suppress.txt &&
+	env LSAN_OPTIONS=suppressions=${CMAKE_CURRENT_BINARY_DIR}/lsan-suppress.txt ${CMAKE_CTEST_COMMAND}
 )
-set_tests_properties(gen_lsan_suppress PROPERTIES FIXTURES_SETUP f_lsan_suppress)
-set_property(TEST PROPERTY FIXTURES_REQUIRED f_lsan_suppress)
 # make -j16 -C ../../build test_heap && ../../build/src/test/test_heap
 # make -j16 -C ../../build test_heap && rm -f $(find ../../build -name '*.gcda') && ctest -V -T test -T coverage -R heap --test-dir ../../build && (cd ../../build; gcovr -f ../src --html --html-nested -o coverage/index.html; cd ../src/test)
 # make -j16 -C ../../build test_blockstore && rm -f $(find ../../build -name '*.gcda') && ctest -V -T test -T coverage -R blockstore --test-dir ../../build && (cd ../../build; gcovr -f ../src --html --html-nested -o coverage/index.html; cd ../src/test)
 # kcov --include-path=../../../src ../../kcov ./test_blockstore
+add_dependencies(test build_tests)
 add_subdirectory(src)

README-ru.md

@@ -26,15 +26,11 @@ Vitastor поддерживает QEMU-драйвер, протоколы UBLK,
 
 ## Презентации и записи докладов
 
-- KuberConf'2025: [видео](https://vitastor.io/presentation/kuberconf.webm)
-- Highload'2025: [видео](https://vitastor.io/presentation/hl2025/hl2025.webm),
-  [на youtube](https://www.youtube.com/watch?v=0R8MLjFtz7g), презентация
-  ([на русском](https://vitastor.io/presentation/hl2025/), [на английском](https://vitastor.io/presentation/hl2025/en.html))
-- Highload'2022: презентация ([на русском](https://vitastor.io/presentation/highload/highload.html)),
-  [видео](https://vitastor.io/presentation/highload/talk.webm)
 - DevOpsConf'2021: презентация ([на русском](https://vitastor.io/presentation/devopsconf/devopsconf.html),
   [на английском](https://vitastor.io/presentation/devopsconf/devopsconf_en.html)),
   [видео](https://vitastor.io/presentation/devopsconf/talk.webm)
+- Highload'2022: презентация ([на русском](https://vitastor.io/presentation/highload/highload.html)),
+  [видео](https://vitastor.io/presentation/highload/talk.webm)
 
 ## Документация
 

README.md

@@ -26,15 +26,11 @@ Read more details in the documentation. You can start from here: [Quick Start](d
 
 ## Talks and presentations
 
-- KuberConf'2025: [video](https://vitastor.io/presentation/kuberconf.webm)
-- Highload'2025: [video](https://vitastor.io/presentation/hl2025/hl2025.webm),
-  [youtube](https://www.youtube.com/watch?v=0R8MLjFtz7g), presentation
-  ([in Russian](https://vitastor.io/presentation/hl2025/), [in English](https://vitastor.io/presentation/hl2025/en.html))
-- Highload'2022: presentation ([in Russian](https://vitastor.io/presentation/highload/highload.html)),
-  [video](https://vitastor.io/presentation/highload/talk.webm)
 - DevOpsConf'2021: presentation ([in Russian](https://vitastor.io/presentation/devopsconf/devopsconf.html),
   [in English](https://vitastor.io/presentation/devopsconf/devopsconf_en.html)),
   [video](https://vitastor.io/presentation/devopsconf/talk.webm)
+- Highload'2022: presentation ([in Russian](https://vitastor.io/presentation/highload/highload.html)),
+  [video](https://vitastor.io/presentation/highload/talk.webm)
 
 ## Documentation
 

csi/Dockerfile

@@ -1,5 +1,5 @@
 # Compile stage
-FROM golang:trixie AS build
+FROM golang:bookworm AS build
 
 ADD go.sum go.mod /app/
 RUN cd /app; CGO_ENABLED=1 GOOS=linux GOARCH=amd64 go mod download -x
@@ -9,7 +9,7 @@ RUN perl -i -e '$/ = undef; while(<>) { s/\n\s*(\{\s*\n)/$1\n/g; s/\}(\s*\n\s*)e
     CGO_ENABLED=1 GOOS=linux GOARCH=amd64 go build -o vitastor-csi
 
 # Final stage
-FROM debian:trixie
+FROM debian:bookworm
 
 LABEL maintainers="Vitaliy Filippov <vitalif@yourcmc.ru>"
 LABEL description="Vitastor CSI Driver"
@@ -25,20 +25,20 @@ RUN apt-get update && \
         # NFS mount dependencies
         nfs-common netbase \
         # dependencies of qemu-storage-daemon
-        libaio1t64 libc6 libfuse3-4 libglib2.0-0t64 libgmp10 libgnutls30t64 \
-        libhogweed6t64 libnettle8t64 libnuma1 libselinux1 liburing2 libzstd1 zlib1g && \
+        libnuma1 liburing2 libglib2.0-0 libfuse3-3 libaio1 libzstd1 libnettle8 \
+        libgmp10 libhogweed6 libp11-kit0 libidn2-0 libunistring2 libtasn1-6 libpcre2-8-0 libffi8 && \
     apt-get clean && \
     (echo options nbd nbds_max=128 > /etc/modprobe.d/nbd.conf)
 
 COPY --from=build /app/vitastor-csi /bin/
 
-RUN (echo deb http://vitastor.io/debian trixie main > /etc/apt/sources.list.d/vitastor.list) && \
+RUN (echo deb http://vitastor.io/debian bookworm main > /etc/apt/sources.list.d/vitastor.list) && \
     ((echo 'Package: *'; echo 'Pin: origin "vitastor.io"'; echo 'Pin-Priority: 1000') > /etc/apt/preferences.d/vitastor.pref) && \
     wget -q -O /etc/apt/trusted.gpg.d/vitastor.gpg https://vitastor.io/debian/pubkey.gpg && \
     apt-get update && \
     apt-get install -y vitastor-client ibverbs-providers && \
-    wget https://vitastor.io/archive/qemu/qemu-trixie-10.0.2%2Bds-2%2Bvitastor1/qemu-utils_10.0.2%2Bds-2%2Bvitastor1_amd64.deb && \
-    wget https://vitastor.io/archive/qemu/qemu-trixie-10.0.2%2Bds-2%2Bvitastor1/qemu-block-extra_10.0.2%2Bds-2%2Bvitastor1_amd64.deb && \
+    wget https://vitastor.io/archive/qemu/qemu-bookworm-9.2.2%2Bds-1%2Bvitastor4/qemu-utils_9.2.2%2Bds-1%2Bvitastor4_amd64.deb && \
+    wget https://vitastor.io/archive/qemu/qemu-bookworm-9.2.2%2Bds-1%2Bvitastor4/qemu-block-extra_9.2.2%2Bds-1%2Bvitastor4_amd64.deb && \
     dpkg -x qemu-utils*.deb tmp1 && \
     dpkg -x qemu-block-extra*.deb tmp1 && \
     cp -a tmp1/usr/bin/qemu-storage-daemon /usr/bin/ && \

csi/Dockerfile.local

@@ -1,5 +1,5 @@
 # Compile stage
-FROM golang:trixie AS build
+FROM golang:bookworm AS build
 
 ADD go.sum go.mod /app/
 RUN cd /app; CGO_ENABLED=1 GOOS=linux GOARCH=amd64 go mod download -x
@@ -9,7 +9,7 @@ RUN perl -i -e '$/ = undef; while(<>) { s/\n\s*(\{\s*\n)/$1\n/g; s/\}(\s*\n\s*)e
     CGO_ENABLED=1 GOOS=linux GOARCH=amd64 go build -o vitastor-csi
 
 # Final stage
-FROM debian:trixie
+FROM debian:bookworm
 
 LABEL maintainers="Vitaliy Filippov <vitalif@yourcmc.ru>"
 LABEL description="Vitastor CSI Driver"
@@ -36,8 +36,8 @@ ADD deb /deb
 
 RUN apt-get update && \
     apt-get -y install /deb/vitastor-client_*.deb && \
-    wget https://vitastor.io/archive/qemu/qemu-trixie-9.2.2%2Bds-1%2Bvitastor4/qemu-utils_9.2.2%2Bds-1%2Bvitastor4_amd64.deb && \
-    wget https://vitastor.io/archive/qemu/qemu-trixie-9.2.2%2Bds-1%2Bvitastor4/qemu-block-extra_9.2.2%2Bds-1%2Bvitastor4_amd64.deb && \
+    wget https://vitastor.io/archive/qemu/qemu-bookworm-9.2.2%2Bds-1%2Bvitastor4/qemu-utils_9.2.2%2Bds-1%2Bvitastor4_amd64.deb && \
+    wget https://vitastor.io/archive/qemu/qemu-bookworm-9.2.2%2Bds-1%2Bvitastor4/qemu-block-extra_9.2.2%2Bds-1%2Bvitastor4_amd64.deb && \
     dpkg -x qemu-utils*.deb tmp1 && \
     dpkg -x qemu-block-extra*.deb tmp1 && \
     cp -a tmp1/usr/bin/qemu-storage-daemon /usr/bin/ && \

csi/Makefile

@@ -1,4 +1,4 @@
-VITASTOR_VERSION ?= v3.0.12
+VITASTOR_VERSION ?= v2.4.3
 
 all: build push
 

csi/deploy/004-csi-nodeplugin.yaml

@@ -49,7 +49,7 @@ spec:
             capabilities:
               add: ["SYS_ADMIN"]
             allowPrivilegeEscalation: true
-          image: vitalif/vitastor-csi:v3.0.12
+          image: vitalif/vitastor-csi:v2.4.3
           args:
             - "--node=$(NODE_ID)"
             - "--endpoint=$(CSI_ENDPOINT)"

csi/deploy/007-csi-provisioner.yaml

@@ -121,7 +121,7 @@ spec:
             privileged: true
             capabilities:
               add: ["SYS_ADMIN"]
-          image: vitalif/vitastor-csi:v3.0.12
+          image: vitalif/vitastor-csi:v2.4.3
           args:
             - "--node=$(NODE_ID)"
             - "--endpoint=$(CSI_ENDPOINT)"

csi/src/config.go

@@ -5,7 +5,7 @@ package vitastor
 
 const (
     vitastorCSIDriverName    = "csi.vitastor.io"
-    vitastorCSIDriverVersion = "3.0.12"
+    vitastorCSIDriverVersion = "2.4.3"
 )
 
 // Config struct fills the parameters of request or user input

debian/build-vitastor-ubuntu-resolute.sh

@@ -1,5 +0,0 @@
-#!/bin/bash
-# 26.04 Resolute Raccoon
-
-docker build --build-arg DISTRO=ubuntu --build-arg REL=resolute -t vitastor-buildenv:resolute -f vitastor-buildenv.Dockerfile .
-docker run -it --rm -e REL=resolute -v `dirname $0`/../:/root/vitastor vitastor-buildenv:resolute /root/vitastor/debian/vitastor-build.sh

debian/changelog

@@ -1,4 +1,4 @@
-vitastor (3.0.12-1) unstable; urgency=medium
+vitastor (2.4.3-1) unstable; urgency=medium
 
   * Bugfixes
 

debian/vitastor-build.sh

@@ -44,7 +44,7 @@ curl -s https://git.yourcmc.ru/vitalif/antietcd/archive/master.tar.gz | tar -zx
 curl -s https://git.yourcmc.ru/vitalif/tinyraft/archive/master.tar.gz | tar -zx
 
 cd /root/vitastor/packages/vitastor-$REL
-if [[ ( "$REL" = "trixie" || "$REL" = "resolute" ) && -e ../vitastor-bookworm/vitastor_$VER.orig.tar.xz ]]; then
+if [[ "$REL" = "trixie" && -e ../vitastor-bookworm/vitastor_$VER.orig.tar.xz ]]; then
     # Fucking shit, archives differ between bookworm (xz 5.4.1) and trixie (xz 5.8.1)
     cp ../vitastor-bookworm/vitastor_$VER.orig.tar.xz .
 else

docker/Dockerfile

@@ -1,6 +1,6 @@
 # Build Docker image with Vitastor packages
 
-FROM debian:trixie
+FROM debian:bookworm
 
 ADD etc/apt /etc/apt/
 RUN apt-get update && apt-get -y install vitastor ibverbs-providers udev systemd qemu-system-x86 qemu-system-common qemu-block-extra qemu-utils jq nfs-common && apt-get clean

docker/Makefile

@@ -1,4 +1,4 @@
-VITASTOR_VERSION ?= v3.0.12
+VITASTOR_VERSION ?= v2.4.3
 
 all: build push
 

docker/etc/apt/preferences

@@ -1,3 +1,3 @@
 Package: *
-Pin: release n=trixie-backports
+Pin: release n=bookworm-backports
 Pin-Priority: 500

docker/etc/apt/sources.list.d/vitastor.list

@@ -1,2 +1,2 @@
-deb http://vitastor.io/debian trixie main
-#deb http://http.debian.net/debian/ trixie-backports main
+deb http://vitastor.io/debian bookworm main
+deb http://http.debian.net/debian/ bookworm-backports main

docker/etc/systemd/system/vitastor-host.service

@@ -7,7 +7,7 @@ PartOf=vitastor.target
 [Service]
 Restart=always
 EnvironmentFile=/etc/vitastor/docker.conf
-ExecStart=bash -c 'docker run --rm -i -v /etc/vitastor:/etc/vitastor -v /dev:/dev -v /run:/run -e SYSTEMD_IN_CHROOT=0 \
+ExecStart=bash -c 'docker run --rm -i -v /etc/vitastor:/etc/vitastor -v /dev:/dev -v /run:/run \
     --security-opt seccomp=unconfined --privileged --pid=host --log-driver none --network host --name vitastor vitastor:$VITASTOR_VERSION \
     sleep.sh'
 ExecStartPost=udevadm trigger

docker/etc/vitastor/docker.conf

@@ -4,7 +4,7 @@
 #
 
 # Desired Vitastor version
-VITASTOR_VERSION=v3.0.12
+VITASTOR_VERSION=v2.4.3
 
 # Additional arguments for all containers
 # For example, you may want to specify a custom logging driver here

docker/install.sh

@@ -2,7 +2,8 @@
 
 set -e
 
-cp -urv /etc/systemd/system/vitastor* /host-etc/systemd/system/
-cp -urv /etc/udev/rules.d /host-etc/udev/
+cp -urv /etc/default /host-etc/
+cp -urv /etc/systemd /host-etc/
+cp -urv /etc/udev /host-etc/
 cp -urnv /etc/vitastor /host-etc/
 cp -urnv /opt/scripts/* /host-bin/

docs/config/layout-osd.en.md

@@ -9,7 +9,6 @@
 These parameters apply to OSDs, are fixed at the moment of OSD drive
 initialization and can't be changed after it without losing data.
 
-- [meta_format](#meta_format)
 - [data_device](#data_device)
 - [meta_device](#meta_device)
 - [journal_device](#journal_device)
@@ -28,21 +27,6 @@ initialization and can't be changed after it without losing data.
 - [data_csum_type](#data_csum_type)
 - [csum_block_size](#csum_block_size)
 
-## meta_format
-
-- Type: integer
-- Default: 3
-
-OSD store implementation version and on-disk metadata format.
-
-Three versions are currently supported: 3, 2 and 1.
-- 3 the new log-structured store, it's overall faster, has lower Write
-  Amplification, which may be even close to 1 (i.e. almost no extra writes)
-  if your SSDs support atomic writes (see [atomic_write_size](osd.en.md#atomic_write_size)).
-- 2 is the old stable store from Vitastor 0.9-2.x.
-- 1 is the same old store but with a legacy metadata format from Vitastor
-  versions to up 0.8.x, without any support for checksums.
-
 ## data_device
 
 - Type: string

docs/config/layout-osd.ru.md

@@ -10,7 +10,6 @@
 дисковые параметры, задаются в момент инициализации дисков OSD и не могут быть
 изменены после этого без потери данных.
 
-- [meta_format](#meta_format)
 - [data_device](#data_device)
 - [meta_device](#meta_device)
 - [journal_device](#journal_device)
@@ -29,23 +28,6 @@
 - [data_csum_type](#data_csum_type)
 - [csum_block_size](#csum_block_size)
 
-## meta_format
-
-- Тип: целое число
-- Значение по умолчанию: 3
-
-Версия реализации дискового хранилища OSD и дискового формата метаданных.
-
-Поддерживаются три версии: 3, 2 и 1.
-- 3 - новое лог-структурированное хранилище, в целом более быстрое, со
-  сниженным фактором амплификации записи, который может составлять около 1
-  (то есть, практически без лишней служебной записи), если ваши SSD
-  поддерживают атомарную запись (см. [atomic_write_size](osd.ru.md#atomic_write_size)).
-- 2 - старое стабильное хранилище из версий Vitastor 0.9-2.x.
-- 1 - то же самое стабильное хранилище, но с ещё более старым форматом
-  метаданных из версий Vitastor до 0.8.x, без какой-либо поддержки
-  контрольных сумм.
-
 ## data_device
 
 - Тип: строка

docs/config/network.en.md

@@ -22,6 +22,7 @@ between clients, OSDs and etcd.
 - [rdma_max_msg](#rdma_max_msg)
 - [rdma_max_recv](#rdma_max_recv)
 - [rdma_max_send](#rdma_max_send)
+- [rdma_odp](#rdma_odp)
 - [peer_connect_interval](#peer_connect_interval)
 - [peer_connect_timeout](#peer_connect_timeout)
 - [osd_idle_timeout](#osd_idle_timeout)
@@ -101,6 +102,11 @@ found or if `osd_network` is not specified. Auto-selection is also
 unsupported with old libibverbs < v32, like in Debian 10 Buster or
 CentOS 7.
 
+Vitastor supports all adapters, even ones without ODP support, like
+Mellanox ConnectX-3 and non-Mellanox cards. Versions up to Vitastor
+1.2.0 required ODP which is only present in Mellanox ConnectX >= 4.
+See also [rdma_odp](#rdma_odp).
+
 Run `ibv_devinfo -v` as root to list available RDMA devices and their
 features.
 
@@ -110,23 +116,6 @@ the manual of your network vendor for details about setting up the switch
 for RoCEv2 correctly. Usually it means setting up Lossless Ethernet with
 PFC (Priority Flow Control) and ECN (Explicit Congestion Notification).
 
-Vitastor supports all adapters, even ones without ODP (On-Demand Paging)
-support, like Mellanox ConnectX-3 and non-Mellanox cards. ODP is only present
-in Mellanox ConnectX >= 4 adapters and allows to skip memory registration
-for RDMA and thus, in theory, avoid memory copying.
-
-Versions up to Vitastor 1.2.0 required ODP, then it was disabled by default,
-but it was still supported up to 3.0.3. Now ODP support is removed because it
-actually only hurts performance: an example 3-node cluster with 8 NVMe in each
-node and 2*25 GBit/s ConnectX-6 RDMA network pushed 3950000 read iops without
-ODP, but only 239000 iops with ODP.
-
-This happens because Mellanox ODP implementation seems to be based on
-message retransmissions when the adapter doesn't know about the buffer yet -
-it likely uses standard "RNR retransmissions" (RNR = receiver not ready)
-which is generally slow in RDMA/RoCE networks. Here's a presentation about
-it from ISPASS-2021 conference: https://tkygtr6.github.io/pub/ISPASS21_slides.pdf
-
 ## rdma_port_num
 
 - Type: integer
@@ -198,6 +187,28 @@ less than `rdma_max_recv` so the receiving side doesn't run out of buffers.
 Doesn't affect memory usage - additional memory isn't allocated for send
 operations.
 
+## rdma_odp
+
+- Type: boolean
+- Default: false
+
+Use RDMA with On-Demand Paging. ODP is currently only available on Mellanox
+ConnectX-4 and newer adapters. ODP allows to not register memory explicitly
+for RDMA adapter to be able to use it. This, in turn, allows to skip memory
+copying during sending. One would think this should improve performance, but
+**in reality** RDMA performance with ODP is **drastically** worse. Example
+3-node cluster with 8 NVMe in each node and 2*25 GBit/s ConnectX-6 RDMA network
+without ODP pushes 3950000 read iops, but only 239000 iops with ODP...
+
+This happens because Mellanox ODP implementation seems to be based on
+message retransmissions when the adapter doesn't know about the buffer yet -
+it likely uses standard "RNR retransmissions" (RNR = receiver not ready)
+which is generally slow in RDMA/RoCE networks. Here's a presentation about
+it from ISPASS-2021 conference: https://tkygtr6.github.io/pub/ISPASS21_slides.pdf
+
+ODP support is retained in the code just in case a good ODP implementation
+appears one day.
+
 ## peer_connect_interval
 
 - Type: seconds

docs/config/network.ru.md

@@ -22,6 +22,7 @@
 - [rdma_max_msg](#rdma_max_msg)
 - [rdma_max_recv](#rdma_max_recv)
 - [rdma_max_send](#rdma_max_send)
+- [rdma_odp](#rdma_odp)
 - [peer_connect_interval](#peer_connect_interval)
 - [peer_connect_timeout](#peer_connect_timeout)
 - [osd_idle_timeout](#osd_idle_timeout)
@@ -100,6 +101,12 @@ RoCEv1/RoCEv2, и даже позволяет полностью отключи
 не задана. Также автовыбор не поддерживается со старыми версиями библиотеки
 libibverbs < v32, например в Debian 10 Buster или CentOS 7.
 
+Vitastor поддерживает все модели адаптеров, включая те, у которых
+нет поддержки ODP, то есть вы можете использовать RDMA с ConnectX-3 и
+картами производства не Mellanox. Версии Vitastor до 1.2.0 включительно
+требовали ODP, который есть только на Mellanox ConnectX 4 и более новых.
+См. также [rdma_odp](#rdma_odp).
+
 Запустите `ibv_devinfo -v` от имени суперпользователя, чтобы посмотреть
 список доступных RDMA-устройств, их параметры и возможности.
 
@@ -110,24 +117,6 @@ libibverbs < v32, например в Debian 10 Buster или CentOS 7.
 подразумевает настройку сети без потерь на основе PFC (Priority Flow
 Control) и ECN (Explicit Congestion Notification).
 
-Vitastor поддерживает все модели адаптеров, включая те, у которых нет
-поддержки ODP (On-Demand Paging), например, ConnectX-3 и карты производства
-не Mellanox. Функция ODP доступна только на адаптерах Mellanox ConnectX-4 и
-более новых и позволяет не регистрировать память для её использования RDMA-картой,
-благодаря чему в теории можно избежать лишних копирований памяти.
-
-Версии Vitastor до 1.2.0 включительно требовали ODP, потом функция был отключена
-по умолчанию, но поддерживалась вплоть до версии 3.0.3. Сейчас поддержка ODP
-полностью удалена, так как на самом деле она только портит производительность:
-например, на 3-узловом кластере с 8 NVMe в каждом узле и сетью 2*25 Гбит/с на
-чтение с RDMA без ODP удаётся снять 3950000 iops, а с ODP - всего 239000 iops.
-
-Это происходит из-за того, что реализация ODP у Mellanox неоптимальная и
-основана на повторной передаче сообщений, когда карте не известен буфер -
-вероятно, на стандартных "RNR retransmission" (RNR = receiver not ready).
-А данные повторные передачи в RDMA/RoCE - всегда очень медленная штука.
-Презентация на эту тему с конференции ISPASS-2021: https://tkygtr6.github.io/pub/ISPASS21_slides.pdf
-
 ## rdma_port_num
 
 - Тип: целое число
@@ -203,6 +192,29 @@ OSD в любом случае согласовывают реальное зн
 Не влияет на потребление памяти - дополнительная память на операции отправки
 не выделяется.
 
+## rdma_odp
+
+- Тип: булево (да/нет)
+- Значение по умолчанию: false
+
+Использовать RDMA с On-Demand Paging. ODP - функция, доступная пока что
+исключительно на адаптерах Mellanox ConnectX-4 и более новых. ODP позволяет
+не регистрировать память для её использования RDMA-картой. Благодаря этому
+можно не копировать данные при отправке их в сеть и, казалось бы, это должно
+улучшать производительность - но **по факту** получается так, что
+производительность только ухудшается, причём сильно. Пример - на 3-узловом
+кластере с 8 NVMe в каждом узле и сетью 2*25 Гбит/с на чтение с RDMA без ODP
+удаётся снять 3950000 iops, а с ODP - всего 239000 iops...
+
+Это происходит из-за того, что реализация ODP у Mellanox неоптимальная и
+основана на повторной передаче сообщений, когда карте не известен буфер -
+вероятно, на стандартных "RNR retransmission" (RNR = receiver not ready).
+А данные повторные передачи в RDMA/RoCE - всегда очень медленная штука.
+Презентация на эту тему с конференции ISPASS-2021: https://tkygtr6.github.io/pub/ISPASS21_slides.pdf
+
+Возможность использования ODP сохранена в коде на случай, если вдруг в один
+прекрасный день появится хорошая реализация ODP.
+
 ## peer_connect_interval
 
 - Тип: секунды

docs/config/osd.en.md

@@ -38,7 +38,6 @@ with an OSD restart or, for some of them, even without restarting by updating co
 - [journal_io](#journal_io)
 - [journal_sector_buffer_count](#journal_sector_buffer_count)
 - [journal_no_same_sector_overwrites](#journal_no_same_sector_overwrites)
-- [skip_corrupted_meta_entries](#skip_corrupted_meta_entries)
 - [throttle_small_writes](#throttle_small_writes)
 - [throttle_target_iops](#throttle_target_iops)
 - [throttle_target_mbs](#throttle_target_mbs)
@@ -66,11 +65,6 @@ with an OSD restart or, for some of them, even without restarting by updating co
 - [allow_net_split](#allow_net_split)
 - [enable_pg_locks](#enable_pg_locks)
 - [pg_lock_retry_interval_ms](#pg_lock_retry_interval_ms)
-- [atomic_write_size](#atomic_write_size)
-- [use_atomic_flag](#use_atomic_flag)
-- [pg_reshard_chunk_size](#pg_reshard_chunk_size)
-- [pg_reshard_chunk_pause_ms](#pg_reshard_chunk_pause_ms)
-- [gc_on_start](#gc_on_start)
 
 ## bind_address
 
@@ -281,19 +275,13 @@ Maximum number of journal flushers (see above min_flusher_count).
 - Type: boolean
 - Default: true
 
-Only for the old store ([meta_format](layout-osd.en.md#meta_format) 2).
-
-This parameter makes Vitastor keep a copy of metadata area in memory as it is
-on disk, in addition to the metadata database. When the option is enabled, every
-metadata entry is effectively stored in RAM twice. It's required for good performance
-because it allows to avoid additional read-modify-write cycles during metadata
-modifications. Metadata area size with the old store is roughly 224 MB per 1 TB
-of data. You can turn the option off to reduce memory usage by this value, but
-it will reduce performance.
-
-For the new store ([meta_format](layout-osd.en.md#meta_format) 3), the option
-may be changed in the future to support operation without loading full metadata
-database in memory.
+This parameter makes Vitastor always keep metadata area of the block device
+in memory. It's required for good performance because it allows to avoid
+additional read-modify-write cycles during metadata modifications. Metadata
+area size is currently roughly 224 MB per 1 TB of data. You can turn it off
+to reduce memory usage by this value, but it will hurt performance. This
+restriction is likely to be removed in the future along with the upgrade
+of the metadata storage scheme.
 
 ## inmemory_journal
 
@@ -372,8 +360,6 @@ blocks. The only situation when you should increase it to a larger value
 is when you enable journal_no_same_sector_overwrites. In this case set
 it to, for example, 1024.
 
-Not applicable to the new store ([meta_format](layout-osd.en.md#meta_format) 3).
-
 ## journal_no_same_sector_overwrites
 
 - Type: boolean
@@ -387,17 +373,6 @@ journal after writing it instead of possibly overwriting it the second time.
 
 Most (99%) other SSDs don't need this option.
 
-Not applicable to the new store ([meta_format](layout-osd.en.md#meta_format) 3).
-
-## skip_corrupted_meta_entries
-
-- Type: boolean
-- Default: false
-
-Only for the new store ([meta_format](layout-osd.en.md#meta_format) 3).
-Allow OSD to start when some metadata entries or blocks are corrupted by
-skipping them. Should be only used as an emergency measure.
-
 ## throttle_small_writes
 
 - Type: boolean
@@ -691,72 +666,3 @@ Use this parameter to enable or disable this function for all pools.
 - Default: 100
 
 Retry interval for failed PG lock attempts.
-
-## atomic_write_size
-
-- Type: integer
-- Default: 4096
-
-Maximum data device atomic write size allowed for OSD to use.
-
-Atomic writes allow to reduce the Write Amplification factor with the new store
-([meta_format](layout-osd.en.md#meta_format)=3) to almost 1 (i.e. almost no extra writes)
-with replicated pools and reach the best possible write performance.
-
-Default value is auto-detected during OSD initialization from
-`/sys/block/xx/queue/atomic_write_max_bytes` or assumed to be 4096 bytes
-because all known disks support 4 KB atomic writes. Auto-detection is only used for
-NVMe disks because SAS disks require the explicit WRITE ATOMIC command which requires
-RWF_ATOMIC (see below [#use_atomic_flag]) but that flag works incorrectly in current
-Linux versions.
-
-You can also check if your NVMe drives support atomic writes by running
-the command `nvme id-ctrl /dev/nvme0n1 | grep awupf`. If the reported value,
-plus 1, multiplied by the currently selected block size of the NVMe,
-is more than 4 KB, then the new store can utilize it for better performance.
-The only drives known to support it currently are [Micron and Kioxia](../intro/quickstart.en.md).
-
-Atomic writes allow to skip double data writes in replicated pools, thus
-reducing Write Amplification and improving write performance up to 2 times.
-
-## use_atomic_flag
-
-- Type: boolean
-
-This option controls whether Vitastor OSDs use RWF_ATOMIC write flag with atomic writes.
-This flag is supported since Linux 6.11 and adds some safety to atomic writes - the kernel
-guarantees to not fragment write requests with it and also to check them against the actual
-device atomic write capabilities.
-
-However, the option is disabled by default because the flag is currently UNUSABLE - Linux
-incorrectly requires writes with that flag to be of power-of-2 length and length-aligned.
-I.e., for example, 12 KB writes and not-8-KB aligned 8 KB writes are forbidden by the kernel,
-even though the NVMe specification allows them.
-
-For NVMe disks with `scheduler=none` writes aren't fragmented anyway so it's not a big deal.
-However, you can rebuild your kernel with [this patch](../../patches/linux-fix-atomic-write-checks.diff)
-and turn this option on. It will make your atomic writes a bit safer.
-
-## pg_reshard_chunk_size
-
-- Type: integer
-- Default: 100000
-
-Pool PG count change is a CPU-intensive operation because OSDs store the full object database
-in memory and have to move all entries between old and new PGs. Thus it's performed in chunks,
-with pauses between chunks to prevent blocking OSD's event loop and other clients' operations.
-This option sets the maximum number of object is a chunk. Moving 100k objects usually takes
-50-100ms. Chunk size equal to 0 means unlimited.
-
-## pg_reshard_chunk_pause_ms
-
-- Type: milliseconds
-- Default: 100
-
-This option sets the interval between handling two PG count change chunks.
-
-## gc_on_start
-
-- Type: boolean
-
-Forcibly clean all garbage entries in the new store on every OSD restart.

docs/config/osd.ru.md

@@ -39,7 +39,6 @@
 - [journal_io](#journal_io)
 - [journal_sector_buffer_count](#journal_sector_buffer_count)
 - [journal_no_same_sector_overwrites](#journal_no_same_sector_overwrites)
-- [skip_corrupted_meta_entries](#skip_corrupted_meta_entries)
 - [throttle_small_writes](#throttle_small_writes)
 - [throttle_target_iops](#throttle_target_iops)
 - [throttle_target_mbs](#throttle_target_mbs)
@@ -67,11 +66,6 @@
 - [allow_net_split](#allow_net_split)
 - [enable_pg_locks](#enable_pg_locks)
 - [pg_lock_retry_interval_ms](#pg_lock_retry_interval_ms)
-- [atomic_write_size](#atomic_write_size)
-- [use_atomic_flag](#use_atomic_flag)
-- [pg_reshard_chunk_size](#pg_reshard_chunk_size)
-- [pg_reshard_chunk_pause_ms](#pg_reshard_chunk_pause_ms)
-- [gc_on_start](#gc_on_start)
 
 ## bind_address
 
@@ -289,19 +283,13 @@ Flusher - это микро-поток (корутина), которая коп
 - Тип: булево (да/нет)
 - Значение по умолчанию: true
 
-Только для старого хранилища ([meta_format](layout-osd.en.md#meta_format) 2).
-
-Данный параметр заставляет Vitastor всегда держать копию области метаданных
-в памяти в том же виде, как она лежит на диске, в дополнение к БД метаданных.
-То есть, с включённой опцией каждая запись метаданных хранится в памяти дважды.
-Это нужно, чтобы избегать дополнительных операций чтения с диска при записи.
-Размер области метаданных в старом хранилище составляет примерно 224 МБ на
-1 ТБ данных. Вы можете отключить опцию, чтобы снизить потребление памяти
-примерно на эту величину, но при этом также снизится и производительность.
-
-Для нового хранилища ([meta_format](layout-osd.en.md#meta_format) 3) опция,
-возможно, будет переработана в будущем для поддержки работы без полной
-загрузки метаданных в памяти.
+Данный параметр заставляет Vitastor всегда держать область метаданных диска
+в памяти. Это нужно, чтобы избегать дополнительных операций чтения с диска
+при записи. Размер области метаданных на данный момент составляет примерно
+224 МБ на 1 ТБ данных. При включении потребление памяти снизится примерно
+на эту величину, но при этом также снизится и производительность. В будущем,
+после обновления схемы хранения метаданных, это ограничение, скорее всего,
+будет ликвидировано.
 
 ## inmemory_journal
 
@@ -384,8 +372,6 @@ fsync небезопасным даже с режимом "directsync".
 нужно менять - это если вы включаете journal_no_same_sector_overwrites. В
 этом случае установите данный параметр, например, в 1024.
 
-Неприменимо к новому хранилищу ([meta_format](layout-osd.en.md#meta_format) 3).
-
 ## journal_no_same_sector_overwrites
 
 - Тип: булево (да/нет)
@@ -401,18 +387,6 @@ fsync небезопасным даже с режимом "directsync".
 
 Почти все другие SSD (99% моделей) не требуют данной опции.
 
-Неприменимо к новому хранилищу ([meta_format](layout-osd.en.md#meta_format) 3).
-
-## skip_corrupted_meta_entries
-
-- Тип: булево (да/нет)
-- Значение по умолчанию: false
-
-Только для нового хранилища ([meta_format](layout-osd.en.md#meta_format) 3).
-Разрешить OSD запускаться, даже если часть блоков или записей метаданных
-повреждена, пропуская их. Опция предназначена для использования только в
-целях аварийного восстановления.
-
 ## throttle_small_writes
 
 - Тип: булево (да/нет)
@@ -725,78 +699,3 @@ pg_minsize OSD во время переключений, что может по
 - Значение по умолчанию: 100
 
 Интервал повтора неудачных попыток блокировки PG.
-
-## atomic_write_size
-
-- Тип: целое число
-- Значение по умолчанию: 4096
-
-Максимальный размер атомарной записи на диск данных, который OSD разрешено использовать.
-
-Поддержка атомарной записи позволяет снизить мультипликатор записи (Write Amplification)
-на диск с новым хранилищем ([meta_format](layout-osd.ru.md#meta_format)=3)
-практически до 1 (то есть, почти до нулевого объёма лишней записи) в реплицированных
-пулах и достигнуть наилучшей возможной производительности записи.
-
-Значение по умолчанию авто-определяется во время инициализации OSD из
-`/sys/block/xx/queue/atomic_write_max_bytes` либо принимается равным 4096,
-так как все известные диски поддерживают атомарную запись 4 КБ блоков.
-Автоопределение применяется только для NVMe-дисков, так как SAS диски требуют
-использования отдельной команды WRITE ATOMIC, а для неё нужен флаг RWF_ATOMIC
-(см. ниже [#use_atomic_flag]), а он в текущих версиях Linux работает некорректно.
-
-Вы также можете проверить, поддерживают ли ваши NVMe-диски атомарную запись,
-с помощью команды `nvme id-ctrl /dev/nvme0n1 | grep awupf`. Если значение awupf
-плюс 1, умноженное на текущий выбранный размер блока NVMe-диска, больше 4 КБ,
-то новое хранилище может использовать атомарные записи для достижения лучшей
-производительности. Единственные известные диски, которые поддерживают это сейчас -
-[Micron и Kioxia](../intro/quickstart.ru.md).
-
-Атомарная запись позволяет не использовать двойную запись данных (в журнал и на
-устройство данных) в реплицированных пулах и таким образом снижает амплификацию
-записи (объём служебной записи на диск) и улучшает производительность записи
-вплоть до 2-х кратного прироста.
-
-## use_atomic_flag
-
-- Тип: булево (да/нет)
-
-Данная опция контролирует использование Vitastor OSD флага RWF_ATOMIC при атомарной записи
-блоков. Этот флаг поддерживается, начиная с версии ядра Linux 6.11 и добавляет немного корректности
-атомарным записям - ядро гарантирует отсутствие фрагментации запросов записи с этим флагом и
-проверяет их на соответствие реальным возможностям устройства.
-
-Однако, данная опция по умолчанию отключена, так как флаг в текущих версиях Linux работает
-абсолютно НЕКОРРЕКТНО - при нём Linux требует, чтобы запросы записи имели длину, равную
-степени двойки и были выровнены на эту длину. То есть, например, 12 КБ запросы записи, а также
-8 КБ запросы записи по не-кратному 8 КБ смещению запрещаются ядром, хотя спецификация NVMe их
-разрешает.
-
-Для NVMe-дисков с `scheduler=none` запросы записи и так не фрагментируются, так что это не так
-уж и важно, однако вы можете пересобрать своё ядро с [этим патчем](../../patches/linux-fix-atomic-write-checks.diff)
-и включить данную опцию. Это сделает вашу атомарную запись капельку безопаснее.
-
-## pg_reshard_chunk_size
-
-- Тип: целое число
-- Значение по умолчанию: 100000
-
-Изменение числа PG в пуле заметно загружает процессор, так как OSD хранят полную базу данных
-объектов в памяти и им приходится перемещать все записи объектов между старыми и новыми PG.
-Поэтому изменение применяется порциями, с паузами между порциями, чтобы не блокировать обработку
-событий OSD и операции остальных клиентов. Данная опция задаёт максимальное число объектов
-в порции. Перемещение 100 тысяч объектов (значение по умолчанию) обычно занимает порядка
-50-100 миллисекунд. Значение опции 0 отключает лимит размера порции.
-
-## pg_reshard_chunk_pause_ms
-
-- Тип: миллисекунды
-- Значение по умолчанию: 100
-
-Данная опция задаёт интервал между обработкой двух порций изменения числа PG пулов.
-
-## gc_on_start
-
-- Тип: булево (да/нет)
-
-Принудительно очищать все мусорные записи в новом хранилище при каждом запуске OSD.

docs/config/src/layout-osd.yml

@@ -1,28 +1,3 @@
-- name: meta_format
-  type: int
-  default: 3
-  info: |
-    OSD store implementation version and on-disk metadata format.
-
-    Three versions are currently supported: 3, 2 and 1.
-    - 3 the new log-structured store, it's overall faster, has lower Write
-      Amplification, which may be even close to 1 (i.e. almost no extra writes)
-      if your SSDs support atomic writes (see [atomic_write_size](osd.en.md#atomic_write_size)).
-    - 2 is the old stable store from Vitastor 0.9-2.x.
-    - 1 is the same old store but with a legacy metadata format from Vitastor
-      versions to up 0.8.x, without any support for checksums.
-  info_ru: |
-    Версия реализации дискового хранилища OSD и дискового формата метаданных.
-
-    Поддерживаются три версии: 3, 2 и 1.
-    - 3 - новое лог-структурированное хранилище, в целом более быстрое, со
-      сниженным фактором амплификации записи, который может составлять около 1
-      (то есть, практически без лишней служебной записи), если ваши SSD
-      поддерживают атомарную запись (см. [atomic_write_size](osd.ru.md#atomic_write_size)).
-    - 2 - старое стабильное хранилище из версий Vitastor 0.9-2.x.
-    - 1 - то же самое стабильное хранилище, но с ещё более старым форматом
-      метаданных из версий Vitastor до 0.8.x, без какой-либо поддержки
-      контрольных сумм.
 - name: data_device
   type: string
   info: |

docs/config/src/network.yml

@@ -84,6 +84,11 @@
     unsupported with old libibverbs < v32, like in Debian 10 Buster or
     CentOS 7.
 
+    Vitastor supports all adapters, even ones without ODP support, like
+    Mellanox ConnectX-3 and non-Mellanox cards. Versions up to Vitastor
+    1.2.0 required ODP which is only present in Mellanox ConnectX >= 4.
+    See also [rdma_odp](#rdma_odp).
+
     Run `ibv_devinfo -v` as root to list available RDMA devices and their
     features.
 
@@ -92,23 +97,6 @@
     the manual of your network vendor for details about setting up the switch
     for RoCEv2 correctly. Usually it means setting up Lossless Ethernet with
     PFC (Priority Flow Control) and ECN (Explicit Congestion Notification).
-
-    Vitastor supports all adapters, even ones without ODP (On-Demand Paging)
-    support, like Mellanox ConnectX-3 and non-Mellanox cards. ODP is only present
-    in Mellanox ConnectX >= 4 adapters and allows to skip memory registration
-    for RDMA and thus, in theory, avoid memory copying.
-
-    Versions up to Vitastor 1.2.0 required ODP, then it was disabled by default,
-    but it was still supported up to 3.0.3. Now ODP support is removed because it
-    actually only hurts performance: an example 3-node cluster with 8 NVMe in each
-    node and 2*25 GBit/s ConnectX-6 RDMA network pushed 3950000 read iops without
-    ODP, but only 239000 iops with ODP.
-
-    This happens because Mellanox ODP implementation seems to be based on
-    message retransmissions when the adapter doesn't know about the buffer yet -
-    it likely uses standard "RNR retransmissions" (RNR = receiver not ready)
-    which is generally slow in RDMA/RoCE networks. Here's a presentation about
-    it from ISPASS-2021 conference: https://tkygtr6.github.io/pub/ISPASS21_slides.pdf
   info_ru: |
     Название RDMA-устройства для связи с Vitastor OSD (например, "rocep5s0f0").
     Если не указано, Vitastor попробует найти RoCE-устройство, соответствующее
@@ -117,6 +105,12 @@
     не задана. Также автовыбор не поддерживается со старыми версиями библиотеки
     libibverbs < v32, например в Debian 10 Buster или CentOS 7.
 
+    Vitastor поддерживает все модели адаптеров, включая те, у которых
+    нет поддержки ODP, то есть вы можете использовать RDMA с ConnectX-3 и
+    картами производства не Mellanox. Версии Vitastor до 1.2.0 включительно
+    требовали ODP, который есть только на Mellanox ConnectX 4 и более новых.
+    См. также [rdma_odp](#rdma_odp).
+
     Запустите `ibv_devinfo -v` от имени суперпользователя, чтобы посмотреть
     список доступных RDMA-устройств, их параметры и возможности.
 
@@ -126,24 +120,6 @@
     коммутатора для RoCEv2 ищите в документации производителя. Обычно это
     подразумевает настройку сети без потерь на основе PFC (Priority Flow
     Control) и ECN (Explicit Congestion Notification).
-
-    Vitastor поддерживает все модели адаптеров, включая те, у которых нет
-    поддержки ODP (On-Demand Paging), например, ConnectX-3 и карты производства
-    не Mellanox. Функция ODP доступна только на адаптерах Mellanox ConnectX-4 и
-    более новых и позволяет не регистрировать память для её использования RDMA-картой,
-    благодаря чему в теории можно избежать лишних копирований памяти.
-
-    Версии Vitastor до 1.2.0 включительно требовали ODP, потом функция был отключена
-    по умолчанию, но поддерживалась вплоть до версии 3.0.3. Сейчас поддержка ODP
-    полностью удалена, так как на самом деле она только портит производительность:
-    например, на 3-узловом кластере с 8 NVMe в каждом узле и сетью 2*25 Гбит/с на
-    чтение с RDMA без ODP удаётся снять 3950000 iops, а с ODP - всего 239000 iops.
-
-    Это происходит из-за того, что реализация ODP у Mellanox неоптимальная и
-    основана на повторной передаче сообщений, когда карте не известен буфер -
-    вероятно, на стандартных "RNR retransmission" (RNR = receiver not ready).
-    А данные повторные передачи в RDMA/RoCE - всегда очень медленная штука.
-    Презентация на эту тему с конференции ISPASS-2021: https://tkygtr6.github.io/pub/ISPASS21_slides.pdf
 - name: rdma_port_num
   type: int
   info: |
@@ -242,6 +218,45 @@
     у принимающей стороны в процессе работы не заканчивались буферы на приём.
     Не влияет на потребление памяти - дополнительная память на операции отправки
     не выделяется.
+- name: rdma_odp
+  type: bool
+  default: false
+  online: false
+  info: |
+    Use RDMA with On-Demand Paging. ODP is currently only available on Mellanox
+    ConnectX-4 and newer adapters. ODP allows to not register memory explicitly
+    for RDMA adapter to be able to use it. This, in turn, allows to skip memory
+    copying during sending. One would think this should improve performance, but
+    **in reality** RDMA performance with ODP is **drastically** worse. Example
+    3-node cluster with 8 NVMe in each node and 2*25 GBit/s ConnectX-6 RDMA network
+    without ODP pushes 3950000 read iops, but only 239000 iops with ODP...
+
+    This happens because Mellanox ODP implementation seems to be based on
+    message retransmissions when the adapter doesn't know about the buffer yet -
+    it likely uses standard "RNR retransmissions" (RNR = receiver not ready)
+    which is generally slow in RDMA/RoCE networks. Here's a presentation about
+    it from ISPASS-2021 conference: https://tkygtr6.github.io/pub/ISPASS21_slides.pdf
+
+    ODP support is retained in the code just in case a good ODP implementation
+    appears one day.
+  info_ru: |
+    Использовать RDMA с On-Demand Paging. ODP - функция, доступная пока что
+    исключительно на адаптерах Mellanox ConnectX-4 и более новых. ODP позволяет
+    не регистрировать память для её использования RDMA-картой. Благодаря этому
+    можно не копировать данные при отправке их в сеть и, казалось бы, это должно
+    улучшать производительность - но **по факту** получается так, что
+    производительность только ухудшается, причём сильно. Пример - на 3-узловом
+    кластере с 8 NVMe в каждом узле и сетью 2*25 Гбит/с на чтение с RDMA без ODP
+    удаётся снять 3950000 iops, а с ODP - всего 239000 iops...
+
+    Это происходит из-за того, что реализация ODP у Mellanox неоптимальная и
+    основана на повторной передаче сообщений, когда карте не известен буфер -
+    вероятно, на стандартных "RNR retransmission" (RNR = receiver not ready).
+    А данные повторные передачи в RDMA/RoCE - всегда очень медленная штука.
+    Презентация на эту тему с конференции ISPASS-2021: https://tkygtr6.github.io/pub/ISPASS21_slides.pdf
+
+    Возможность использования ODP сохранена в коде на случай, если вдруг в один
+    прекрасный день появится хорошая реализация ODP.
 - name: peer_connect_interval
   type: sec
   min: 1

docs/config/src/osd.yml

@@ -253,33 +253,21 @@
   type: bool
   default: true
   info: |
-    Only for the old store ([meta_format](layout-osd.en.md#meta_format) 2).
-
-    This parameter makes Vitastor keep a copy of metadata area in memory as it is
-    on disk, in addition to the metadata database. When the option is enabled, every
-    metadata entry is effectively stored in RAM twice. It's required for good performance
-    because it allows to avoid additional read-modify-write cycles during metadata
-    modifications. Metadata area size with the old store is roughly 224 MB per 1 TB
-    of data. You can turn the option off to reduce memory usage by this value, but
-    it will reduce performance.
-
-    For the new store ([meta_format](layout-osd.en.md#meta_format) 3), the option
-    may be changed in the future to support operation without loading full metadata
-    database in memory.
+    This parameter makes Vitastor always keep metadata area of the block device
+    in memory. It's required for good performance because it allows to avoid
+    additional read-modify-write cycles during metadata modifications. Metadata
+    area size is currently roughly 224 MB per 1 TB of data. You can turn it off
+    to reduce memory usage by this value, but it will hurt performance. This
+    restriction is likely to be removed in the future along with the upgrade
+    of the metadata storage scheme.
   info_ru: |
-    Только для старого хранилища ([meta_format](layout-osd.en.md#meta_format) 2).
-
-    Данный параметр заставляет Vitastor всегда держать копию области метаданных
-    в памяти в том же виде, как она лежит на диске, в дополнение к БД метаданных.
-    То есть, с включённой опцией каждая запись метаданных хранится в памяти дважды.
-    Это нужно, чтобы избегать дополнительных операций чтения с диска при записи.
-    Размер области метаданных в старом хранилище составляет примерно 224 МБ на
-    1 ТБ данных. Вы можете отключить опцию, чтобы снизить потребление памяти
-    примерно на эту величину, но при этом также снизится и производительность.
-
-    Для нового хранилища ([meta_format](layout-osd.en.md#meta_format) 3) опция,
-    возможно, будет переработана в будущем для поддержки работы без полной
-    загрузки метаданных в памяти.
+    Данный параметр заставляет Vitastor всегда держать область метаданных диска
+    в памяти. Это нужно, чтобы избегать дополнительных операций чтения с диска
+    при записи. Размер области метаданных на данный момент составляет примерно
+    224 МБ на 1 ТБ данных. При включении потребление памяти снизится примерно
+    на эту величину, но при этом также снизится и производительность. В будущем,
+    после обновления схемы хранения метаданных, это ограничение, скорее всего,
+    будет ликвидировано.
 - name: inmemory_journal
   type: bool
   default: true
@@ -398,15 +386,11 @@
     blocks. The only situation when you should increase it to a larger value
     is when you enable journal_no_same_sector_overwrites. In this case set
     it to, for example, 1024.
-
-    Not applicable to the new store ([meta_format](layout-osd.en.md#meta_format) 3).
   info_ru: |
     Максимальное число буферов, разрешённых для использования под записываемые
     в журнал блоки метаданных. Единственная ситуация, в которой этот параметр
     нужно менять - это если вы включаете journal_no_same_sector_overwrites. В
     этом случае установите данный параметр, например, в 1024.
-
-    Неприменимо к новому хранилищу ([meta_format](layout-osd.en.md#meta_format) 3).
 - name: journal_no_same_sector_overwrites
   type: bool
   default: false
@@ -418,8 +402,6 @@
     journal after writing it instead of possibly overwriting it the second time.
 
     Most (99%) other SSDs don't need this option.
-
-    Not applicable to the new store ([meta_format](layout-osd.en.md#meta_format) 3).
   info_ru: |
     Включайте данную опцию для SSD вроде Intel D3-S4510 и D3-S4610, которые
     ОЧЕНЬ не любят, когда ПО перезаписывает один и тот же сектор несколько раз
@@ -430,20 +412,6 @@
     самого сектора.
 
     Почти все другие SSD (99% моделей) не требуют данной опции.
-
-    Неприменимо к новому хранилищу ([meta_format](layout-osd.en.md#meta_format) 3).
-- name: skip_corrupted_meta_entries
-  type: bool
-  default: false
-  info: |
-    Only for the new store ([meta_format](layout-osd.en.md#meta_format) 3).
-    Allow OSD to start when some metadata entries or blocks are corrupted by
-    skipping them. Should be only used as an emergency measure.
-  info_ru: |
-    Только для нового хранилища ([meta_format](layout-osd.en.md#meta_format) 3).
-    Разрешить OSD запускаться, даже если часть блоков или записей метаданных
-    повреждена, пропуская их. Опция предназначена для использования только в
-    целях аварийного восстановления.
 - name: throttle_small_writes
   type: bool
   default: false
@@ -833,112 +801,3 @@
   default: 100
   info: Retry interval for failed PG lock attempts.
   info_ru: Интервал повтора неудачных попыток блокировки PG.
-- name: atomic_write_size
-  type: int
-  default: 4096
-  info: |
-    Maximum data device atomic write size allowed for OSD to use.
-
-    Atomic writes allow to reduce the Write Amplification factor with the new store
-    ([meta_format](layout-osd.en.md#meta_format)=3) to almost 1 (i.e. almost no extra writes)
-    with replicated pools and reach the best possible write performance.
-
-    Default value is auto-detected during OSD initialization from
-    `/sys/block/xx/queue/atomic_write_max_bytes` or assumed to be 4096 bytes
-    because all known disks support 4 KB atomic writes. Auto-detection is only used for
-    NVMe disks because SAS disks require the explicit WRITE ATOMIC command which requires
-    RWF_ATOMIC (see below [#use_atomic_flag]) but that flag works incorrectly in current
-    Linux versions.
-
-    You can also check if your NVMe drives support atomic writes by running
-    the command `nvme id-ctrl /dev/nvme0n1 | grep awupf`. If the reported value,
-    plus 1, multiplied by the currently selected block size of the NVMe,
-    is more than 4 KB, then the new store can utilize it for better performance.
-    The only drives known to support it currently are [Micron and Kioxia](../intro/quickstart.en.md).
-
-    Atomic writes allow to skip double data writes in replicated pools, thus
-    reducing Write Amplification and improving write performance up to 2 times.
-  info_ru: |
-    Максимальный размер атомарной записи на диск данных, который OSD разрешено использовать.
-
-    Поддержка атомарной записи позволяет снизить мультипликатор записи (Write Amplification)
-    на диск с новым хранилищем ([meta_format](layout-osd.ru.md#meta_format)=3)
-    практически до 1 (то есть, почти до нулевого объёма лишней записи) в реплицированных
-    пулах и достигнуть наилучшей возможной производительности записи.
-
-    Значение по умолчанию авто-определяется во время инициализации OSD из
-    `/sys/block/xx/queue/atomic_write_max_bytes` либо принимается равным 4096,
-    так как все известные диски поддерживают атомарную запись 4 КБ блоков.
-    Автоопределение применяется только для NVMe-дисков, так как SAS диски требуют
-    использования отдельной команды WRITE ATOMIC, а для неё нужен флаг RWF_ATOMIC
-    (см. ниже [#use_atomic_flag]), а он в текущих версиях Linux работает некорректно.
-
-    Вы также можете проверить, поддерживают ли ваши NVMe-диски атомарную запись,
-    с помощью команды `nvme id-ctrl /dev/nvme0n1 | grep awupf`. Если значение awupf
-    плюс 1, умноженное на текущий выбранный размер блока NVMe-диска, больше 4 КБ,
-    то новое хранилище может использовать атомарные записи для достижения лучшей
-    производительности. Единственные известные диски, которые поддерживают это сейчас -
-    [Micron и Kioxia](../intro/quickstart.ru.md).
-
-    Атомарная запись позволяет не использовать двойную запись данных (в журнал и на
-    устройство данных) в реплицированных пулах и таким образом снижает амплификацию
-    записи (объём служебной записи на диск) и улучшает производительность записи
-    вплоть до 2-х кратного прироста.
-- name: use_atomic_flag
-  type: bool
-  info: |
-    This option controls whether Vitastor OSDs use RWF_ATOMIC write flag with atomic writes.
-    This flag is supported since Linux 6.11 and adds some safety to atomic writes - the kernel
-    guarantees to not fragment write requests with it and also to check them against the actual
-    device atomic write capabilities.
-
-    However, the option is disabled by default because the flag is currently UNUSABLE - Linux
-    incorrectly requires writes with that flag to be of power-of-2 length and length-aligned.
-    I.e., for example, 12 KB writes and not-8-KB aligned 8 KB writes are forbidden by the kernel,
-    even though the NVMe specification allows them.
-
-    For NVMe disks with `scheduler=none` writes aren't fragmented anyway so it's not a big deal.
-    However, you can rebuild your kernel with [this patch](../../patches/linux-fix-atomic-write-checks.diff)
-    and turn this option on. It will make your atomic writes a bit safer.
-  info_ru: |
-    Данная опция контролирует использование Vitastor OSD флага RWF_ATOMIC при атомарной записи
-    блоков. Этот флаг поддерживается, начиная с версии ядра Linux 6.11 и добавляет немного корректности
-    атомарным записям - ядро гарантирует отсутствие фрагментации запросов записи с этим флагом и
-    проверяет их на соответствие реальным возможностям устройства.
-
-    Однако, данная опция по умолчанию отключена, так как флаг в текущих версиях Linux работает
-    абсолютно НЕКОРРЕКТНО - при нём Linux требует, чтобы запросы записи имели длину, равную
-    степени двойки и были выровнены на эту длину. То есть, например, 12 КБ запросы записи, а также
-    8 КБ запросы записи по не-кратному 8 КБ смещению запрещаются ядром, хотя спецификация NVMe их
-    разрешает.
-
-    Для NVMe-дисков с `scheduler=none` запросы записи и так не фрагментируются, так что это не так
-    уж и важно, однако вы можете пересобрать своё ядро с [этим патчем](../../patches/linux-fix-atomic-write-checks.diff)
-    и включить данную опцию. Это сделает вашу атомарную запись капельку безопаснее.
-- name: pg_reshard_chunk_size
-  type: int
-  default: 100000
-  info: |
-    Pool PG count change is a CPU-intensive operation because OSDs store the full object database
-    in memory and have to move all entries between old and new PGs. Thus it's performed in chunks,
-    with pauses between chunks to prevent blocking OSD's event loop and other clients' operations.
-    This option sets the maximum number of object is a chunk. Moving 100k objects usually takes
-    50-100ms. Chunk size equal to 0 means unlimited.
-  info_ru: |
-    Изменение числа PG в пуле заметно загружает процессор, так как OSD хранят полную базу данных
-    объектов в памяти и им приходится перемещать все записи объектов между старыми и новыми PG.
-    Поэтому изменение применяется порциями, с паузами между порциями, чтобы не блокировать обработку
-    событий OSD и операции остальных клиентов. Данная опция задаёт максимальное число объектов
-    в порции. Перемещение 100 тысяч объектов (значение по умолчанию) обычно занимает порядка
-    50-100 миллисекунд. Значение опции 0 отключает лимит размера порции.
-- name: pg_reshard_chunk_pause_ms
-  type: ms
-  default: 100
-  info: |
-    This option sets the interval between handling two PG count change chunks.
-  info_ru: |
-    Данная опция задаёт интервал между обработкой двух порций изменения числа PG пулов.
-- name: gc_on_start
-  type: bool
-  info: Forcibly clean all garbage entries in the new store on every OSD restart.
-  info_ru: Принудительно очищать все мусорные записи в новом хранилище при каждом запуске OSD.

docs/installation/docker.en.md

@@ -26,37 +26,13 @@ at Vitastor Kubernetes operator: https://github.com/Antilles7227/vitastor-operat
 The instruction is very simple.
 
 1. Download a Docker image of the desired version: \
-   `docker pull vitalif/vitastor:v3.0.12`
+   `docker pull vitalif/vitastor:v2.4.3`
 2. Install scripts to the host system: \
-   `docker run --rm -it -v /etc:/host-etc -v /usr/bin:/host-bin vitalif/vitastor:v3.0.12 install.sh`
+   `docker run --rm -it -v /etc:/host-etc -v /usr/bin:/host-bin vitalif/vitastor:v2.4.3 install.sh`
 3. Reload udev rules: \
    `udevadm control --reload-rules`
-4. Enable the vitastor-host service: \
-   `systemctl enable --now vitastor-host`
 
-After these steps, you can return to [Quick Start](../intro/quickstart.en.md).
-
-## Podman
-
-If you use Podman, run the following commands as root before installing Vitastor containers:
-
-```
-ln -s podman /usr/bin/docker
-
-mkdir -p /etc/systemd/system/systemd-udevd.service.d
-
-cat >/etc/systemd/system/systemd-udevd.service.d/override.conf <<EOF
-[Service]
-CapabilityBoundingSet=~
-SystemCallFilter=@mount capset
-EOF
-
-systemctl daemon-reload
-
-systemctl restart systemd-udevd
-```
-
-Without it, udev fails to do calls into a Podman container and Vitastor disk detection doesn't work.
+And you can return to [Quick Start](../intro/quickstart.en.md).
 
 ## Upgrading Containers
 

docs/installation/docker.ru.md

@@ -25,39 +25,14 @@ Vitastor можно установить в Docker/Podman. При этом etcd,
 Инструкция по установке максимально простая.
 
 1. Скачайте Docker-образ желаемой версии: \
-   `docker pull vitalif/vitastor:v3.0.12`
+   `docker pull vitalif/vitastor:v2.4.3`
 2. Установите скрипты в хост-систему командой: \
-   `docker run --rm -it -v /etc:/host-etc -v /usr/bin:/host-bin vitalif/vitastor:v3.0.12 install.sh`
+   `docker run --rm -it -v /etc:/host-etc -v /usr/bin:/host-bin vitalif/vitastor:v2.4.3 install.sh`
 3. Перезагрузите правила udev: \
    `udevadm control --reload-rules`
-4. Включите сервис vitastor-host: \
-   `systemctl enable --now vitastor-host`
 
 После этого вы можете возвращаться к разделу [Быстрый старт](../intro/quickstart.ru.md).
 
-## Podman
-
-Если вы используете Podman, перед установкой контейнеров Vitastor выполните следующие
-команды от имени суперпользователя:
-
-```
-ln -s podman /usr/bin/docker
-
-mkdir -p /etc/systemd/system/systemd-udevd.service.d
-
-cat >/etc/systemd/system/systemd-udevd.service.d/override.conf <<EOF
-[Service]
-CapabilityBoundingSet=~
-SystemCallFilter=@mount capset
-EOF
-
-systemctl daemon-reload
-
-systemctl restart systemd-udevd
-```
-
-Без этих настроек udev не может делать вызовы внутрь Podman-контейнеров и определение дисков Vitastor не работает.
-
 ## Обновление контейнеров
 
 Сначала обязательно проверьте раздел [Обновление Vitastor](../usage/admin.ru.md#обновление-vitastor),

docs/installation/packages.en.md

@@ -17,7 +17,6 @@
   - Debian 10 (Buster): `deb https://vitastor.io/debian buster main`
   - Ubuntu 22.04 (Jammy): `deb https://vitastor.io/debian jammy main`
   - Ubuntu 24.04 (Noble): `deb https://vitastor.io/debian noble main`
-  - Ubuntu 26.04 (Resolute): `deb https://vitastor.io/debian resolute main`
   - Add `-oldstable` to bookworm/bullseye/buster in this line to install the last
     stable version from 0.9.x branch instead of 1.x
 - To always prefer vitastor-patched QEMU and Libvirt versions, add the following to `/etc/apt/preferences`:
@@ -34,17 +33,15 @@
   - CentOS 7: `yum install https://vitastor.io/rpms/centos/7/vitastor-release.rpm`
   - CentOS 8: `dnf install https://vitastor.io/rpms/centos/8/vitastor-release.rpm`
   - AlmaLinux 9 and other RHEL 9 clones (Rocky, Oracle...): `dnf install https://vitastor.io/rpms/centos/9/vitastor-release.rpm`
-  - AlmaLinux 10 and other RHEL 10 clones: `dnf install https://vitastor.io/rpms/centos/10/vitastor-release.rpm`
 - Enable EPEL: `yum/dnf install epel-release`
 - Enable additional CentOS repositories:
   - CentOS 7: `yum install centos-release-scl`
   - CentOS 8: `dnf install centos-release-advanced-virtualization`
-  - RHEL 9/10 clones: not required
+  - RHEL 9 clones: not required
 - Enable elrepo-kernel:
   - CentOS 7: `yum install https://www.elrepo.org/elrepo-release-7.el7.elrepo.noarch.rpm`
   - CentOS 8: `dnf install https://www.elrepo.org/elrepo-release-8.el8.elrepo.noarch.rpm`
   - RHEL 9 clones: `dnf install https://www.elrepo.org/elrepo-release-9.el9.elrepo.noarch.rpm`
-  - RHEL 10 clones: not required
 - Install packages: `yum/dnf install vitastor lpsolve etcd kernel-ml qemu-kvm`
 
 ## Installation requirements

docs/installation/packages.ru.md

@@ -17,7 +17,6 @@
   - Debian 10 (Buster): `deb https://vitastor.io/debian buster main`
   - Ubuntu 22.04 (Jammy): `deb https://vitastor.io/debian jammy main`
   - Ubuntu 24.04 (Noble): `deb https://vitastor.io/debian noble main`
-  - Ubuntu 26.04 (Resolute): `deb https://vitastor.io/debian resolute main`
   - Добавьте `-oldstable` к слову bookworm/bullseye/buster в этой строке, чтобы
     установить последнюю стабильную версию из ветки 0.9.x вместо 1.x
 - Чтобы всегда предпочитались версии пакетов QEMU и Libvirt с патчами Vitastor, добавьте в `/etc/apt/preferences`:
@@ -34,17 +33,15 @@
   - CentOS 7: `yum install https://vitastor.io/rpms/centos/7/vitastor-release.rpm`
   - CentOS 8: `dnf install https://vitastor.io/rpms/centos/8/vitastor-release.rpm`
   - AlmaLinux 9 и другие клоны RHEL 9 (Rocky, Oracle...): `dnf install https://vitastor.io/rpms/centos/9/vitastor-release.rpm`
-  - AlmaLinux 10 и другие клоны RHEL 10: `dnf install https://vitastor.io/rpms/centos/10/vitastor-release.rpm`
 - Включите EPEL: `yum/dnf install epel-release`
 - Включите дополнительные репозитории CentOS:
   - CentOS 7: `yum install centos-release-scl`
   - CentOS 8: `dnf install centos-release-advanced-virtualization`
-  - Клоны RHEL 9/10: не нужно
+  - Клоны RHEL 9: не нужно
 - Включите elrepo-kernel:
   - CentOS 7: `yum install https://www.elrepo.org/elrepo-release-7.el7.elrepo.noarch.rpm`
   - CentOS 8: `dnf install https://www.elrepo.org/elrepo-release-8.el8.elrepo.noarch.rpm`
   - Клоны RHEL 9: `dnf install https://www.elrepo.org/elrepo-release-9.el9.elrepo.noarch.rpm`
-  - Клоны RHEL 10: не нужно
 - Установите пакеты: `yum/dnf install vitastor lpsolve etcd kernel-ml qemu-kvm`
 
 ## Установочные требования

docs/installation/proxmox.en.md

@@ -6,7 +6,7 @@
 
 # Proxmox VE
 
-To enable Vitastor support in Proxmox Virtual Environment (6.4-9.x are supported):
+To enable Vitastor support in Proxmox Virtual Environment (6.4-8.x are supported):
 
 - Add the corresponding Vitastor Debian repository into sources.list on Proxmox hosts:
   trixie for 9.0+, bookworm for 8.1+, pve8.0 for 8.0, bullseye for 7.4, pve7.3 for 7.3, pve7.2 for 7.2, pve7.1 for 7.1, buster for 6.4

docs/installation/proxmox.ru.md

@@ -6,7 +6,7 @@
 
 # Proxmox VE
 
-Чтобы подключить Vitastor к Proxmox Virtual Environment (поддерживаются версии 6.4-9.x):
+Чтобы подключить Vitastor к Proxmox Virtual Environment (поддерживаются версии 6.4-8.x):
 
 - Добавьте соответствующий Debian-репозиторий Vitastor в sources.list на хостах Proxmox:
   trixie для 9.0+, bookworm для 8.1+, pve8.0 для 8.0, bullseye для 7.4, pve7.3 для 7.3, pve7.2 для 7.2, pve7.1 для 7.1, buster для 6.4

docs/installation/source.en.md

@@ -16,7 +16,8 @@
   designated initializers support from C++20
 - CMake
 - jerasure headers and libraries
-- ISA-L, libibverbs, librdmacm, libnl3 headers and libraries (optional)
+- ISA-L, libibverbs and librdmacm headers and libraries (optional)
+- tcmalloc (google-perftools-dev)
 
 ## Basic instructions
 

docs/installation/source.ru.md

@@ -16,7 +16,8 @@
   назначенных инициализаторов (designated initializers) из C++20
 - CMake
 - Заголовки и библиотеки jerasure
-- Опционально - заголовки и библиотеки ISA-L, libibverbs, librdmacm, libnl3
+- Опционально - заголовки и библиотеки ISA-L, libibverbs, librdmacm
+- tcmalloc (google-perftools-dev)
 
 ## Базовая инструкция
 

docs/intro/features.en.md

@@ -14,8 +14,6 @@
 
 - Basic part: highly-available block storage with symmetric clustering and no SPOF
 - [Performance](../performance/bench2.en.md) ;-D
-  - [NVMe atomic write support](../config/osd.en.md#atomic_write_size) for reducing the amount
-    of "extra" disk writes to almost zero (Write Amplification = 1)
 - [Multiple redundancy schemes](../config/pool.en.md#scheme): Replication, XOR n+1, Reed-Solomon erasure codes
   based on jerasure and ISA-L libraries with any number of data and parity drives in a group
 - Configuration via simple JSON data structures in etcd (parameters, pools and images)

docs/intro/features.ru.md

@@ -14,8 +14,6 @@
 
 - Базовая часть - надёжное кластерное блочное хранилище без единой точки отказа
 - [Производительность](../performance/bench2.ru.md) ;-D
-  - [Поддержка атомарной записи NVMe](../config/osd.ru.md#atomic_write_size) для снижения объёма
-    служебной записи практически до нуля (Write Amplification = 1)
 - [Несколько схем отказоустойчивости](../config/pool.ru.md#scheme): репликация, XOR n+1 (1 диск чётности), коды коррекции ошибок
   Рида-Соломона на основе библиотек jerasure и ISA-L с любым числом дисков данных и чётности в группе
 - Конфигурация через простые человекочитаемые JSON-структуры в etcd

docs/intro/quickstart.en.md

@@ -18,10 +18,9 @@
 
 ## Preparation
 
-- Get some SATA or NVMe SSDs with capacitors (server-grade drives). The best performance
-  is achieved with Micron or Kioxia NVMes with atomic write support (see below). You can use desktop
-  SSDs with lazy fsync, but prepare for inferior single-thread latency. Read more about
-  capacitors [here](../config/layout-cluster.en.md#immediate_commit).
+- Get some SATA or NVMe SSDs with capacitors (server-grade drives). You can use desktop SSDs
+  with lazy fsync, but prepare for inferior single-thread latency. Read more about capacitors
+  [here](../config/layout-cluster.en.md#immediate_commit).
 - If you want to use HDDs, get modern HDDs with Media Cache or SSD Cache: HGST Ultrastar,
   Toshiba MG, Seagate EXOS or something similar. If your drives don't have such cache then
   you also need small SSDs for journal and metadata (even 2 GB per 1 TB of HDD space is enough).
@@ -31,11 +30,9 @@
 
 ## Recommended drives
 
-- NVMe with atomic write support (ideal!): Micron 7450/7500/7550, Kioxia CD6/CD7/CD8/CD9
-- Other NVMe: Micron 9100/9200/9300/9400/9550, Micron 7300, Samsung PM983/PM9A3, Samsung PM1723/1735/1743,
-  Intel DC-P3700/P4500/P4600, Intel/Solidigm D5-P4320/P5530, Intel/Solidigm D7-P5500/P5600, Solidigm D7-PS1010/PS1030/P5810,
-  Intel Optane, Kingston DC1000B/DC1500M, Kioxia CD6/CD7/CD8/CD9
-- SATA SSD: Micron 5100/5200/5300/5400, Samsung PM863/PM883/PM893, Intel/Solidigm D3-S4510/4520/4610/4620, Kingston DC500M
+- SATA SSD: Micron 5100/5200/5300/5400, Samsung PM863/PM883/PM893, Intel D3-S4510/4520/4610/4620, Kingston DC500M
+- NVMe: Micron 9100/9200/9300/9400, Micron 7300/7450, Samsung PM983/PM9A3, Samsung PM1723/1735/1743,
+  Intel DC-P3700/P4500/P4600, Intel D5-P4320/P5530, Intel D7-P5500/P5600, Intel Optane, Kingston DC1000B/DC1500M
 - HDD: HGST Ultrastar, Toshiba MG, Seagate EXOS
 
 ## Configure monitors

docs/intro/quickstart.ru.md

@@ -18,9 +18,8 @@
 
 ## Подготовка
 
-- Возьмите серверы с SSD (SATA или NVMe), желательно с конденсаторами (серверные SSD). Наилучшая
-  производительность достигается на дисках Micron и Kioxia с поддержкой атомарной записи (см. ниже).
-  Можно использовать и десктопные SSD, включив режим отложенного fsync, но производительность будет хуже.
+- Возьмите серверы с SSD (SATA или NVMe), желательно с конденсаторами (серверные SSD). Можно
+  использовать и десктопные SSD, включив режим отложенного fsync, но производительность будет хуже.
   О конденсаторах читайте [здесь](../config/layout-cluster.ru.md#immediate_commit).
 - Если хотите использовать HDD, берите современные модели с Media или SSD кэшем - HGST Ultrastar,
   Toshiba MG, Seagate EXOS или что-то похожее. Если такого кэша у ваших дисков нет,
@@ -31,11 +30,9 @@
 
 ## Рекомендуемые диски
 
-- NVMe с поддержкой атомарной записи (идеально!): Micron 7450/7500/7550, Kioxia CD6/CD7/CD8/CD9
-- Другие NVMe: Micron 9100/9200/9300/9400/9550, Micron 7300, Samsung PM983/PM9A3, Samsung PM1723/1735/1743,
-  Intel DC-P3700/P4500/P4600, Intel/Solidigm D5-P4320/P5530, Intel/Solidigm D7-P5500/P5600, Solidigm D7-PS1010/PS1030/P5810,
-  Intel Optane, Kingston DC1000B/DC1500M, Kioxia CD6/CD7/CD8/CD9
-- SATA SSD: Micron 5100/5200/5300/5400, Samsung PM863/PM883/PM893, Intel/Solidigm D3-S4510/4520/4610/4620, Kingston DC500M
+- SATA SSD: Micron 5100/5200/5300/5400, Samsung PM863/PM883/PM893, Intel D3-S4510/4520/4610/4620, Kingston DC500M
+- NVMe: Micron 9100/9200/9300/9400, Micron 7300/7450, Samsung PM983/PM9A3, Samsung PM1723/1735/1743,
+  Intel DC-P3700/P4500/P4600, Intel D5-P4320/P5530, Intel D7-P5500/P5600, Intel Optane, Kingston DC1000B/DC1500M
 - HDD: HGST Ultrastar, Toshiba MG, Seagate EXOS
 
 ## Настройте мониторы

docs/performance/theoretical.en.md

@@ -14,7 +14,7 @@ Replicated setups:
 - Linear read: `min(total network bandwidth, sum(disk read MB/s))`.
 - Linear write: `min(total network bandwidth, sum(disk write MB/s / number of replicas))`.
 - Saturated parallel read iops: `min(total network bandwidth, sum(disk read iops))`.
-- Saturated parallel write iops: `min(total network bandwidth / number of replicas, sum(disk write iops / number of replicas / write amplification))`.
+- Saturated parallel write iops: `min(total network bandwidth / number of replicas, sum(disk write iops / number of replicas / (write amplification = 4)))`.
 
 EC/XOR setups (EC N+K):
 - Single-threaded (T1Q1) read latency: 1.5 network roundtrips + 1 disk read.
@@ -26,36 +26,28 @@ EC/XOR setups (EC N+K):
 - Linear read: `min(total network bandwidth, sum(disk read MB/s))`.
 - Linear write: `min(total network bandwidth, sum(disk write MB/s * N/(N+K)))`.
 - Saturated parallel read iops: `min(total network bandwidth, sum(disk read iops))`.
-- Saturated parallel write iops: roughly `total iops / (N+K) / WA`. More exactly:
-  - With the new store: `min(total network bandwidth * N/(N+K), sum(disk randrw iops / (2 + N-1 + K*2)))`,
-    with random read/write mix corresponding to `(N-1)/(2 + N-1 + K*2)*100 % reads`.
-    - For example, with EC 2+1 it is: `(20% randrw iops) / 5`.
-    - With EC 6+3 it is: `(38% randrw iops) / 13`.
-  - With the old store: `min(total network bandwidth * N/(N+K), sum(disk randrw iops / (3 + N-1 + K*3)))`,
-    with random read/write mix corresponding to `(N-1)/(3 + N-1 + K*3)*100 % reads`.
-    - For example, with EC 2+1 it is: `(14% randrw iops) / 7`.
-    - With EC 6+3 it is: `(30% randrw iops) / 17`.
+- Saturated parallel write iops: roughly `total iops / (N+K) / WA`. More exactly,
+  `min(total network bandwidth * N/(N+K), sum(disk randrw iops / (N*4 + K*5 + 1)))` with
+  random read/write mix corresponding to `(N-1)/(N*4 + K*5 + 1)*100 % reads`.
+  - For example, with EC 2+1 it is: `(7% randrw iops) / 14`.
+  - With EC 6+3 it is: `(12.5% randrw iops) / 40`.
 
-Write Amplification factor:
-- For the new store and for 4 KB writes: WA is always 1 unless you set [atomic_write_size](../config/osd.en.md#atomic_write_size) to 0 manually.
-- For the new store and for 8-124 KB writes: WA is 1 if you use NVMe drives with atomic write support, or roughly 2 if you use other drives.
-- For the old store, WA is roughly `(2 * write size + 4 KB) / (write size)`. So, for 4 KB writes it's 3, and for 8-124 KB writes it's closer to 2.
-- For both the new and the old store and for writes of [block_size](../config/layout-cluster.en.md#block_size): WA is almost 1.
+Write amplification for 4 KB blocks is usually 3-5 in Vitastor:
+1. Journal block write
+2. Journal data write
+3. Metadata block write
+4. Another journal block write for EC/XOR setups
+5. Data block write
 
-Write Amplification consists of:
-- For the new store:
-  - Buffer block write if non-atomic
-  - Data block write
-  - Metadata write(s) (amortized)
-- For the old store:
-  - Journal block write (amortized)
-  - Journal data write
-  - Metadata block write
-  - Another journal block write for EC/XOR setups (amortized)
-  - Data block write
+If you manage to get an SSD which handles 512 byte blocks well (Optane?) you may
+lower 1, 3 and 4 to 512 bytes (1/8 of data size) and get WA as low as 2.375.
 
-Other possibilities to reduce WA would be to use SSDs with internal 512-byte blocks
-or NVDIMM, but both options seem unavailable on the market at the moment.
+Implemented NVDIMM support can basically eliminate WA at all - all extra writes will
+go to DRAM memory. But this requires a test cluster with NVDIMM - please contact me
+if you want to provide me with such cluster for tests.
+
+Lazy fsync also reduces WA for parallel workloads because journal blocks are only
+written when they fill up or fsync is requested.
 
 ## In Practice
 

docs/performance/theoretical.ru.md

@@ -27,36 +27,29 @@
 - Линейное чтение: сумма МБ/с чтения всех дисков, либо общая производительность сети, если в сеть упрётся раньше.
 - Линейная запись: сумма МБ/с записи всех дисков * N/(N+K), либо производительность сети * N / (N+K), если в сеть упрётся раньше.
 - Параллельное случайное мелкое чтение: сумма IOPS чтения всех дисков либо производительность сети, если в сеть упрётся раньше.
-- Параллельная случайная мелкая запись: грубо `(сумма IOPS / (N+K) / WA)`.
-  Либо `производительность сети * N/(N+K)`, если в сеть упрётся раньше. Если точнее, то:
-  - С новым хранилищем: сумма смешанного IOPS всех дисков при `(N-1)/(2 + N-1 + K*2)*100 %` чтения, делённая на `(2 + N-1 + K*2)`.
-    - Например, при EC 2+1 это: `(сумма IOPS при 20% чтения) / 5`.
-    - При EC 6+3 это: `(сумма IOPS при 38% чтения) / 13`.
-  - Со старым хранилищем: сумма смешанного IOPS всех дисков при `(N-1)/(3 + N-1 + K*3)*100 %` чтения, делённая на `(3 + N-1 + K*3)`.
-    - Например, при EC 2+1 это: `(сумма IOPS при 14% чтения) / 7`.
-    - При EC 6+3 это: `(сумма IOPS при 30% чтения) / 17`.
+- Параллельная случайная мелкая запись: грубо `(сумма IOPS / (N+K) / WA)`. Если точнее, то:
+  сумма смешанного IOPS всех дисков при `(N-1)/(N*4 + K*5 + 1)*100 %` чтения, делённая на `(N*4 + K*5 + 1)`.
+  Либо, производительность сети * N/(N+K), если в сеть упрётся раньше.
+  - Например, при EC 2+1 это: `(сумма IOPS при 7% чтения) / 14`.
+  - При EC 6+3 это: `(сумма IOPS при 12.5% чтения) / 40`.
 
-WA (Write Amplification, мультипликатор записи):
-- С новым хранилищем для 4 КБ записи: WA всегда примерно 1, если только вы не установите [atomic_write_size](../config/osd.ru.md#atomic_write_size) вручную в 0.
-- С новым хранилищем и большими записями (8-124 КБ): WA примерно 1, если вы используете NVMe-диски с поддержкой атомарной записи,
-  или примерно 2, если вы используете другие диски.
-- Со старым хранилищем, WA примерно `(2 * размер записи + 4 КБ) / (размер записи)`. То есть, для 4 КБ записи WA=3, а для 8-124 КБ WA ближе к 2.
-- И с новым, и со старым хранилищем и для записи размером [block_size](../config/layout-cluster.ru.md#block_size): WA примерно равен 1.
+WA (мультипликатор записи) для 4 КБ блоков в Vitastor обычно составляет 3-5:
+1. Запись метаданных в журнал
+2. Запись блока данных в журнал
+3. Запись метаданных в БД
+4. Ещё одна запись метаданных в журнал при использовании EC
+5. Запись блока данных на диск данных
 
-Мультипликатор записи состоит из:
-- С новым хранилищем:
-  - Запись блока буфера, если диски без поддержки атомарной записи
-  - Запись блока данных
-  - Запись(-и) блоков метаданных (амортизированные)
-- Со старым хранилищем:
-  - Запись блока журнала (амортизированная)
-  - Запись данных в журнал
-  - Запись блока метаданных
-  - Ещё одна запись блока журнала для EC/XOR пулов (амортизированная)
-  - Запись блока данных
+Если вы найдёте SSD, хорошо работающий с 512-байтными блоками данных (Optane?),
+то 1, 3 и 4 можно снизить до 512 байт (1/8 от размера данных) и получить WA всего 2.375.
 
-Другими потенциальными возможностями снижения WA могли бы быть SSD с внутренним 512-байтным блоком
-либо NVDIMM, но и то, и другое сейчас выглядит недоступным на рынке.
+Если реализовать поддержку NVDIMM, то WA можно, условно говоря, ликвидировать вообще - все
+дополнительные операции записи смогут обслуживаться DRAM памятью. Но для этого необходим
+тестовый кластер с NVDIMM - пишите, если готовы предоставить такой для тестов.
+
+Кроме того, WA снижается при использовании отложенного/ленивого сброса при параллельной
+нагрузке, т.к. блоки журнала записываются на диск только когда они заполняются или явным
+образом запрашивается fsync.
 
 ## На практике
 

docs/usage/admin.en.md

@@ -231,18 +231,6 @@ Upgrading from <= 0.5.x to >= 0.6.x is not supported.
 
 Downgrade are also allowed freely, except the following specific instructions:
 
-### 3.x -> 2.x
-
-Versions 3.0.0 and newer contain two store implementations - an old one and a new
-one, unsupported in 2.x and previous versions. So you should check your OSD store
-versions before downgrading to 2.x with the following command:
-
-`vitastor-disk read-sb /dev/vitastor/osdXX-data | jq -r .meta_format`
-
-If it prints 3 then OSD uses the new store and you can't downgrade it to 2.x.
-
-If it prints 2 or nothing then OSD uses the old store and the downgrade is allowed.
-
 ### 1.8.0 to 1.7.1
 
 Before downgrading from version >= 1.8.0 to version <= 1.7.1

docs/usage/admin.ru.md

@@ -228,18 +228,6 @@ done
 
 Откат (понижение версии) тоже свободно разрешён, кроме указанных ниже случаев:
 
-### 3.x -> 2.x
-
-Версии 3.0.0 и более новые содержат две реализации хранилища - старую и новую, не
-поддерживаемую в 2.x и предыдущих версиях. Таким образом, перед откатом на 2.x вам
-следует проверить, какая версия хранилища используется вашими OSD - командой:
-
-`vitastor-disk read-sb /dev/vitastor/osdXX-data | jq -r .meta_format`
-
-Если выводится 3, это новое хранилище и откатить такой OSD до 2.x нельзя.
-
-Если выводится 2 или не выводится ничего, это старое хранилище и откат разрешён.
-
 ### 1.8.0 -> 1.7.1
 
 Перед понижением версии с >= 1.8.0 до <= 1.7.1 вы должны скопировать ключ

docs/usage/disk.en.md

@@ -51,9 +51,6 @@ Options (automatic mode):
 ```
 --osd_per_disk <N>
   Create <N> OSDs on each disk (default 1)
---meta_format 3
-  Metadata store version. 3 is the new log-structured store, 2 is the stable store
-  from Vitastor 0.9-2.x, 1 is the legacy store from Vitastor 0.6-0.8.
 --hybrid
   Prepare hybrid (HDD+SSD, NVMe+SATA or etc) OSDs using provided devices. By default,
   any passed SSDs will be used for journals and metadata, HDDs will be used for data,
@@ -95,8 +92,6 @@ Options (single-device mode):
 Options (both modes):
 
 ```
---tags tag1,tag2           Set new OSD tag(s)
---weight <number>          Set new OSD weight (between 0 to 1)
 --journal_size 1G/32M      Set journal size (area or partition size)
 --block_size 1M/128k       Set blockstore object size
 --bitmap_granularity 4k    Set bitmap granularity

docs/usage/disk.ru.md

@@ -50,9 +50,6 @@ vitastor-disk - инструмент командной строки для уп
 ```
 --osd_per_disk <N>
   Создавать по несколько (<N>) OSD на каждом диске (по умолчанию 1)
---meta_format 3
-  Версия хранилища метаданных. 3 - новое лог-структурированное хранилище,
-  2 - стабильное хранилище из Vitastor 0.9-2.x, 1 - старое хранилище из Vitastor 0.6-0.8.
 --hybrid
   Инициализировать гибридные (HDD+SSD, NVMe+SATA и т.п.) OSD на указанных дисках.
   По умолчанию, SSD будут использованы для журналов и метаданных, а HDD - для данных,
@@ -96,8 +93,6 @@ vitastor-disk - инструмент командной строки для уп
 Опции для обоих режимов:
 
 ```
---tags tag1,tag2           Задать теги для новых OSD
---weight <number>          Задать вес для новых OSD (от 0 до 1)
 --journal_size 1G/32M      Задать размер журнала (области или раздела журнала)
 --block_size 1M/128k       Задать размер объекта хранилища
 --bitmap_granularity 4k    Задать гранулярность битовых карт

docs/usage/nfs.en.md

@@ -262,4 +262,3 @@ Options:
 | `--logfile <FILE>` | log to the specified file                                |
 | `--enforce 1`      | enforce permissions at the server side (no by default)   |
 | `--foreground 1`   | stay in foreground, do not daemonize                     |
-| `--trace`          | trace all NFS requests                                   |

docs/usage/nfs.ru.md

@@ -274,4 +274,3 @@ VitastorFS из GPUDirect.
 | `--logfile <FILE>` | записывать логи в заданный файл                         |
 | `--enforce 1`      | проверять права доступа на стороне сервера (по умолчанию нет) |
 | `--foreground 1`   | не уходить в фон после запуска                          |
-| `--trace`          | логгировать все запросы NFS                             |

mon/http_server.js

@@ -16,7 +16,7 @@ async function create_http_server(cfg, handler)
         };
         if (cfg.mon_https_ca)
         {
-            tls.ca = await fsp.readFile(cfg.mon_https_ca);
+            tls.mon_https_ca = await fsp.readFile(cfg.mon_https_ca);
         }
         if (cfg.mon_https_client_auth)
         {

mon/lp_optimizer/lp_optimizer.js

@@ -10,19 +10,16 @@ const NO_OSD = 'Z';
 async function lp_solve(text)
 {
     const cp = child_process.spawn('lp_solve');
-    let stdout = '', stderr = '', finish_cb, finished = 0;
+    let stdout = '', stderr = '', finish_cb;
     cp.stdout.on('data', buf => stdout += buf.toString());
     cp.stderr.on('data', buf => stderr += buf.toString());
-    cp.stdout.on('end', () => finish_cb());
-    cp.stderr.on('end', () => finish_cb());
+    cp.on('exit', () => finish_cb && finish_cb());
     cp.stdin.write(text);
     cp.stdin.end();
-    await new Promise(ok => (finish_cb = () =>
+    if (cp.exitCode == null)
     {
-        finished++;
-        if (finished == 2)
-            ok();
-    }));
+        await new Promise(ok => finish_cb = ok);
+    }
     if (!stdout.trim())
     {
         return null;

mon/mon.js

@@ -627,7 +627,7 @@ class Mon
                 if (this.state.pg.history[pool_id] &&
                     this.state.pg.history[pool_id][pg])
                 {
-                    pg_history[pg-1] = JSON.parse(JSON.stringify(this.state.pg.history[pool_id][pg]));
+                    pg_history[pg-1] = this.state.pg.history[pool_id][pg];
                 }
             }
             const real_prev_pgs = [];
@@ -719,7 +719,7 @@ class Mon
                 this.next_recheck_timer = null;
                 this.next_recheck_at = 0;
                 this.schedule_recheck();
-            }, (this.next_recheck_at-now)*1000);
+            }, now-this.next_recheck_at);
         }
     }
 

mon/osd_tree.js

@@ -87,7 +87,7 @@ function make_hier_tree(global_config, tree)
     tree[''] = { children: [] };
     for (const node_id in tree)
     {
-        if (node_id === '')
+        if (node_id === '' || !(tree[node_id].children||[]).length && (tree[node_id].size||0) <= 0)
         {
             continue;
         }

mon/package.json

@@ -1,6 +1,6 @@
 {
   "name": "vitastor-mon",
-  "version": "3.0.12",
+  "version": "2.4.3",
   "description": "Vitastor SDS monitor service",
   "main": "mon-main.js",
   "scripts": {
@@ -9,7 +9,7 @@
   "author": "Vitaliy Filippov",
   "license": "UNLICENSED",
   "dependencies": {
-    "antietcd": "^1.2.4",
+    "antietcd": "^1.1.3",
     "sprintf-js": "^1.1.2",
     "ws": "^7.2.5"
   },

mon/pg_gen.js

@@ -9,6 +9,7 @@ const LPOptimizer = require('./lp_optimizer/lp_optimizer.js');
 const { scale_pg_count } = require('./pg_utils.js');
 const { make_hier_tree, filter_osds_by_root_node,
     filter_osds_by_tags, filter_osds_by_block_layout, get_affinity_osds } = require('./osd_tree.js');
+const { select_murmur3 } = require('./lp_optimizer/murmur3.js');
 
 function pick_primary(pool_id, pg_num, pool_config, osd_set, up_osds, aff_osds)
 {
@@ -38,7 +39,7 @@ function pick_primary(pool_id, pg_num, pool_config, osd_set, up_osds, aff_osds)
     {
         return 0;
     }
-    return alive_set[pg_num % alive_set.length];
+    return alive_set[select_murmur3(alive_set.length, osd_num => pool_id+'/'+pg_num+'/'+osd_num)];
 }
 
 function recheck_primary(state, global_config, up_osds, osd_tree)
@@ -52,7 +53,6 @@ function recheck_primary(state, global_config, up_osds, osd_tree)
             continue;
         }
         const aff_osds = get_affinity_osds(pool_cfg, up_osds, osd_tree);
-        let paused = false;
         for (let pg_num = 1; pg_num <= pool_cfg.pg_count; pg_num++)
         {
             if (!state.pg.config.items[pool_id])
@@ -75,19 +75,6 @@ function recheck_primary(state, global_config, up_osds, osd_tree)
                     );
                     new_pg_config.items[pool_id][pg_num].primary = new_primary;
                 }
-                paused = paused || !!pg_cfg.pause;
-            }
-        }
-        if (paused)
-        {
-            if (!new_pg_config)
-            {
-                new_pg_config = JSON.parse(JSON.stringify(state.pg.config));
-            }
-            console.log(`Resuming paused pool ${pool_id}`);
-            for (const pg in new_pg_config.items[pool_id])
-            {
-                delete new_pg_config.items[pool_id][pg].pause;
             }
         }
     }
@@ -192,10 +179,10 @@ async function generate_pool_pgs(state, global_config, pool_id, osd_tree, levels
     const rules = use_rules ? get_pg_rules(pool_id, pool_cfg, global_config.placement_levels) : null;
     const folded = fold_failure_domains(Object.values(pool_tree), use_rules ? rules : [ [ [ pool_cfg.failure_domain ] ] ]);
     // FIXME: Remove/merge make_hier_tree() step somewhere, however it's needed to remove empty nodes
-    const folded_tree = make_hier_tree(global_config, folded.nodes.reduce((a, c) => { a[c.id] = c; return a; }, {}));
+    const folded_tree = make_hier_tree(global_config, folded.nodes);
     const old_pg_count = prev_pgs.length;
     const optimize_cfg = {
-        osd_weights: folded.nodes.reduce((a, c) => { if (/^\d+$/.exec(c.id) && c.size != null) { a[c.id] = c.size||0; } return a; }, {}),
+        osd_weights: folded.nodes.reduce((a, c) => { if (Number(c.id)) { a[c.id] = c.size; } return a; }, {}),
         combinator: use_rules
             // new algorithm:
             ? new RuleCombinator(folded_tree, rules, pool_cfg.max_osd_combinations)

mon/pg_utils.js

@@ -84,7 +84,7 @@ function scale_pg_history(prev_pg_history, prev_pgs, new_pgs)
         finish_pg_history(merged_history[1]);
         for (let i = 0; i < new_pg_count; i++)
         {
-            new_pg_history[i] = JSON.parse(JSON.stringify(merged_history[1]));
+            new_pg_history[i] = { ...merged_history[1] };
         }
     }
     // Mark history keys for removed PGs as removed
@@ -102,7 +102,7 @@ function scale_pg_count(prev_pgs, new_pg_count)
     {
         for (let i = prev_pgs.length; i < new_pg_count; i++)
         {
-            prev_pgs[i] = [ ...prev_pgs[i % prev_pgs.length] ];
+            prev_pgs[i] = prev_pgs[i % prev_pgs.length];
         }
     }
     else if (prev_pgs.length > new_pg_count)

mon/stats.js

@@ -37,7 +37,6 @@ function derive_osd_stats(st, prev, prev_diff)
         const n = c.count - BigInt(pr && pr.count||0);
         diff.recovery_stats[op] = { ...c, bps: n > 0 ? b*1000n/timediff : 0n, iops: n > 0 ? n*1000n/timediff : 0n };
     }
-    diff.inode_stats = {};
     for (const pool_id in st.inode_stats||{})
     {
         diff.inode_stats[pool_id] = {};

node-binding/package.json

@@ -1,6 +1,6 @@
 {
   "name": "vitastor",
-  "version": "3.0.12",
+  "version": "2.4.3",
   "description": "Low-level native bindings to Vitastor client library",
   "main": "index.js",
   "keywords": [

patches/cinder-vitastor.py

@@ -50,7 +50,7 @@ from cinder.volume import configuration
 from cinder.volume import driver
 from cinder.volume import volume_utils
 
-VITASTOR_VERSION = '3.0.12'
+VITASTOR_VERSION = '2.4.3'
 
 LOG = logging.getLogger(__name__)
 
@@ -275,7 +275,7 @@ class VitastorDriver(driver.CloneableImageVD,
             LOG.exception('error getting vitastor pool stats: '+str(e))
 
         self._stats = stats
-
+        
     def get_volume_stats(self, refresh=False):
         """Get volume stats.
         If 'refresh' is True, run update the stats first.
@@ -291,14 +291,6 @@ class VitastorDriver(driver.CloneableImageVD,
         else:
             return (1 + resp['kvs'][0]['value'], resp['kvs'][0]['mod_revision'])
 
-    def _cli(self, descr, *args):
-        args = [ 'vitastor-cli', *args, *(self._vitastor_args()) ]
-        try:
-            self._execute(*args)
-        except processutils.ProcessExecutionError as exc:
-            LOG.error("Failed to "+descr+": "+exc)
-            raise exception.VolumeBackendAPIException(data = exc.stderr)
-
     def create_volume(self, volume):
         """Creates a logical volume."""
 
@@ -310,7 +302,7 @@ class VitastorDriver(driver.CloneableImageVD,
 
         LOG.debug("creating volume '%s'", vol_name)
 
-        self._cli('create volume', 'create', vol_name, '--size', size)
+        self._create_image(vol_name, { 'size': size })
 
         if volume.encryption_key_id:
             self._create_encrypted_volume(volume, volume.obj_context)
@@ -354,7 +346,7 @@ class VitastorDriver(driver.CloneableImageVD,
         snap_name = utils.convert_str(snapshot.name)
         if snap_name.find('@') >= 0 or snap_name.find('/') >= 0:
             raise exception.VolumeBackendAPIException(data = '@ and / are forbidden in volume and snapshot names')
-        self._cli('create snapshot', 'snap-create', vol_name+'@'+snap_name)
+        self._create_snapshot(vol_name, vol_name+'@'+snap_name)
 
     def snapshot_revert_use_temp_snapshot(self):
         """Disable the use of a temporary snapshot on revert."""
@@ -367,8 +359,21 @@ class VitastorDriver(driver.CloneableImageVD,
         snap_name = utils.convert_str(snapshot.name)
 
         # Delete the image and recreate it from the snapshot
-        self._cli('delete image', 'rm', vol_name)
-        self._cli('recreate image', 'create', '--parent', vol_name+'@'+snap_name, vol_name)
+        args = [ 'vitastor-cli', 'rm', vol_name, *(self._vitastor_args()) ]
+        try:
+            self._execute(*args)
+        except processutils.ProcessExecutionError as exc:
+            LOG.error("Failed to delete image "+vol_name+": "+exc)
+            raise exception.VolumeBackendAPIException(data = exc.stderr)
+        args = [
+            'vitastor-cli', 'create', '--parent', vol_name+'@'+snap_name,
+            vol_name, *(self._vitastor_args())
+        ]
+        try:
+            self._execute(*args)
+        except processutils.ProcessExecutionError as exc:
+            LOG.error("Failed to recreate image "+vol_name+" from "+vol_name+"@"+snap_name+": "+exc)
+            raise exception.VolumeBackendAPIException(data = exc.stderr)
 
     def delete_snapshot(self, snapshot):
         """Deletes a snapshot."""
@@ -376,7 +381,15 @@ class VitastorDriver(driver.CloneableImageVD,
         vol_name = utils.convert_str(snapshot.volume_name)
         snap_name = utils.convert_str(snapshot.name)
 
-        self._cli('remove snapshot', 'rm', vol_name+'@'+snap_name)
+        args = [
+            'vitastor-cli', 'rm', vol_name+'@'+snap_name,
+            *(self._vitastor_args())
+        ]
+        try:
+            self._execute(*args)
+        except processutils.ProcessExecutionError as exc:
+            LOG.error("Failed to remove snapshot "+vol_name+'@'+snap_name+": "+exc)
+            raise exception.VolumeBackendAPIException(data = exc.stderr)
 
     def _child_count(self, parents):
         children = 0
@@ -414,7 +427,13 @@ class VitastorDriver(driver.CloneableImageVD,
         if src_vref.admin_metadata.get('readonly') == 'True':
             # source volume is a volume-image cache entry or other readonly volume
             # clone without intermediate snapshot
-            self._cli('create clone', 'create', '--parent', src_name, '--size', size, dest_name)
+            src = self._get_image(src_name)
+            LOG.debug("creating image '%s' from '%s'", dest_name, src_name)
+            new_cfg = self._create_image(dest_name, {
+                'size': size,
+                'parent_id': src['idx']['id'],
+                'parent_pool_id': src['idx']['pool_id'],
+            })
             return {}
 
         clone_snap = "%s@%s.clone_snap" % (src_name, dest_name)
@@ -427,12 +446,15 @@ class VitastorDriver(driver.CloneableImageVD,
             clone_snap = dest_name
             make_img = False
 
-        LOG.debug("creating snapshot '%s'", clone_snap)
-        self._cli('create base snapshot', 'snap-create', '--allow-existing', '1', clone_snap)
-
+        LOG.debug("creating layer '%s' under '%s'", clone_snap, src_name)
+        new_cfg = self._create_snapshot(src_name, clone_snap, True)
         if make_img:
             # Then create a clone from it
-            self._cli('create clone', 'create', '--parent', clone_snap, '--size', size, dest_name)
+            new_cfg = self._create_image(dest_name, {
+                'size': size,
+                'parent_id': new_cfg['parent_id'],
+                'parent_pool_id': new_cfg['parent_pool_id'],
+            })
 
         return {}
 
@@ -442,8 +464,7 @@ class VitastorDriver(driver.CloneableImageVD,
         vol_name = utils.convert_str(volume.name)
         snap_name = utils.convert_str(snapshot.name)
 
-        src_snap = 'volume-'+snapshot.volume_id+'@'+snap_name
-        snap = self._get_image(src_snap)
+        snap = self._get_image('volume-'+snapshot.volume_id+'@'+snap_name)
         if not snap:
             raise exception.SnapshotNotFound(snapshot_id = snap_name)
         snap_inode_id = int(resp['responses'][0]['kvs'][0]['value']['id'])
@@ -452,8 +473,12 @@ class VitastorDriver(driver.CloneableImageVD,
         size = snap['cfg']['size']
         if int(volume.size):
             size = int(volume.size) * units.Gi
+        new_cfg = self._create_image(vol_name, {
+            'size': size,
+            'parent_id': snap['idx']['id'],
+            'parent_pool_id': snap['idx']['pool_id'],
+        })
 
-        self._cli('create clone', 'create', vol_name, '--size', size, '--parent', src_snap)
         return {}
 
     def _vitastor_args(self):
@@ -480,7 +505,49 @@ class VitastorDriver(driver.CloneableImageVD,
         """Deletes a logical volume."""
 
         vol_name = utils.convert_str(volume.name)
-        self._cli('delete volume', 'rm', '--matching', vol_name, vol_name+'@*', '--progress', '0')
+
+        # Find the volume and all its snapshots
+        range_end = b'index/image/' + vol_name.encode('utf-8')
+        range_end = range_end[0 : len(range_end)-1] + six.int2byte(range_end[len(range_end)-1] + 1)
+        resp = self._etcd_txn({ 'success': [
+            { 'request_range': { 'key': 'index/image/'+vol_name, 'range_end': range_end } },
+        ] })
+        if len(resp['responses'][0]['kvs']) == 0:
+            # already deleted
+            LOG.info("volume %s no longer exists in backend", vol_name)
+            return
+        layers = resp['responses'][0]['kvs']
+        layer_ids = {}
+        for kv in layers:
+            inode_id = int(kv['value']['id'])
+            pool_id = int(kv['value']['pool_id'])
+            inode_pool_id = (pool_id << 48) | (inode_id & 0xffffffffffff)
+            layer_ids[inode_pool_id] = True
+
+        # Check if the volume has clones and raise 'busy' if so
+        children = self._child_count(layer_ids)
+        if children > 0:
+            raise exception.VolumeIsBusy(volume_name = vol_name)
+
+        # Clear data
+        for kv in layers:
+            args = [
+                'vitastor-cli', 'rm-data', '--pool', str(kv['value']['pool_id']),
+                '--inode', str(kv['value']['id']), '--progress', '0',
+                *(self._vitastor_args())
+            ]
+            try:
+                self._execute(*args)
+            except processutils.ProcessExecutionError as exc:
+                LOG.error("Failed to remove layer "+kv['key']+": "+exc)
+                raise exception.VolumeBackendAPIException(data = exc.stderr)
+
+        # Delete all layers from etcd
+        requests = []
+        for kv in layers:
+            requests.append({ 'request_delete_range': { 'key': kv['key'] } })
+            requests.append({ 'request_delete_range': { 'key': 'config/inode/'+str(kv['value']['pool_id'])+'/'+str(kv['value']['id']) } })
+        self._etcd_txn({ 'success': requests })
 
     def retype(self, context, volume, new_type, diff, host):
         """Change extra type specifications for a volume."""
@@ -500,6 +567,98 @@ class VitastorDriver(driver.CloneableImageVD,
         """Removes an export for a logical volume."""
         pass
 
+    def _create_image(self, vol_name, cfg):
+        pool_s = str(self.cfg['pool_id'])
+        image_id = 0
+        while image_id == 0:
+            # check if the image already exists and find a free ID
+            resp = self._etcd_txn({ 'success': [
+                { 'request_range': { 'key': 'index/image/'+vol_name } },
+                { 'request_range': { 'key': 'index/maxid/'+pool_s } },
+            ] })
+            if len(resp['responses'][0]['kvs']) > 0:
+                # already exists
+                raise exception.VolumeBackendAPIException(data = 'Volume '+vol_name+' already exists')
+            image_id, id_mod = self._next_id(resp['responses'][1])
+            # try to create the image
+            resp = self._etcd_txn({ 'compare': [
+                { 'target': 'MOD', 'mod_revision': id_mod, 'key': 'index/maxid/'+pool_s },
+                { 'target': 'VERSION', 'version': 0, 'key': 'index/image/'+vol_name },
+                { 'target': 'VERSION', 'version': 0, 'key': 'config/inode/'+pool_s+'/'+str(image_id) },
+            ], 'success': [
+                { 'request_put': { 'key': 'index/maxid/'+pool_s, 'value': image_id } },
+                { 'request_put': { 'key': 'index/image/'+vol_name, 'value': json.dumps({
+                    'id': image_id, 'pool_id': self.cfg['pool_id']
+                }) } },
+                { 'request_put': { 'key': 'config/inode/'+pool_s+'/'+str(image_id), 'value': json.dumps({
+                    **cfg, 'name': vol_name,
+                }) } },
+            ] })
+            if not resp.get('succeeded'):
+                # repeat
+                image_id = 0
+
+    def _create_snapshot(self, vol_name, snap_vol_name, allow_existing = False):
+        while True:
+            # check if the image already exists and snapshot doesn't
+            resp = self._etcd_txn({ 'success': [
+                { 'request_range': { 'key': 'index/image/'+vol_name } },
+                { 'request_range': { 'key': 'index/image/'+snap_vol_name } },
+            ] })
+            if len(resp['responses'][0]['kvs']) == 0:
+                raise exception.VolumeBackendAPIException(data = 'Volume '+vol_name+' does not exist')
+            if len(resp['responses'][1]['kvs']) > 0:
+                if allow_existing:
+                    snap_idx = resp['responses'][1]['kvs'][0]['value']
+                    resp = self._etcd_txn({ 'success': [
+                        { 'request_range': { 'key': 'config/inode/'+str(snap_idx['pool_id'])+'/'+str(snap_idx['id']) } },
+                    ] })
+                    if len(resp['responses'][0]['kvs']) == 0:
+                        raise exception.VolumeBackendAPIException(data =
+                            'Volume '+snap_vol_name+' is already indexed, but does not exist'
+                        )
+                    return resp['responses'][0]['kvs'][0]['value']
+                raise exception.VolumeBackendAPIException(
+                    data = 'Volume '+snap_vol_name+' already exists'
+                )
+            vol_idx = resp['responses'][0]['kvs'][0]['value']
+            vol_idx_mod = resp['responses'][0]['kvs'][0]['mod_revision']
+            # get image inode config and find a new ID
+            resp = self._etcd_txn({ 'success': [
+                { 'request_range': { 'key': 'config/inode/'+str(vol_idx['pool_id'])+'/'+str(vol_idx['id']) } },
+                { 'request_range': { 'key': 'index/maxid/'+str(self.cfg['pool_id']) } },
+            ] })
+            if len(resp['responses'][0]['kvs']) == 0:
+                raise exception.VolumeBackendAPIException(data = 'Volume '+vol_name+' does not exist')
+            vol_cfg = resp['responses'][0]['kvs'][0]['value']
+            vol_mod = resp['responses'][0]['kvs'][0]['mod_revision']
+            new_id, id_mod = self._next_id(resp['responses'][1])
+            # try to redirect image to the new inode
+            new_cfg = {
+                **vol_cfg, 'name': vol_name, 'parent_id': vol_idx['id'], 'parent_pool_id': vol_idx['pool_id']
+            }
+            resp = self._etcd_txn({ 'compare': [
+                { 'target': 'MOD', 'mod_revision': vol_idx_mod, 'key': 'index/image/'+vol_name },
+                { 'target': 'MOD', 'mod_revision': vol_mod, 'key': 'config/inode/'+str(vol_idx['pool_id'])+'/'+str(vol_idx['id']) },
+                { 'target': 'MOD', 'mod_revision': id_mod, 'key': 'index/maxid/'+str(self.cfg['pool_id']) },
+                { 'target': 'VERSION', 'version': 0, 'key': 'index/image/'+snap_vol_name },
+                { 'target': 'VERSION', 'version': 0, 'key': 'config/inode/'+str(self.cfg['pool_id'])+'/'+str(new_id) },
+            ], 'success': [
+                { 'request_put': { 'key': 'index/maxid/'+str(self.cfg['pool_id']), 'value': new_id } },
+                { 'request_put': { 'key': 'index/image/'+vol_name, 'value': json.dumps({
+                    'id': new_id, 'pool_id': self.cfg['pool_id']
+                }) } },
+                { 'request_put': { 'key': 'config/inode/'+str(self.cfg['pool_id'])+'/'+str(new_id), 'value': json.dumps(new_cfg) } },
+                { 'request_put': { 'key': 'index/image/'+snap_vol_name, 'value': json.dumps({
+                    'id': vol_idx['id'], 'pool_id': vol_idx['pool_id']
+                }) } },
+                { 'request_put': { 'key': 'config/inode/'+str(vol_idx['pool_id'])+'/'+str(vol_idx['id']), 'value': json.dumps({
+                    **vol_cfg, 'name': snap_vol_name, 'readonly': True
+                }) } }
+            ] })
+            if resp.get('succeeded'):
+                return new_cfg
+
     def initialize_connection(self, volume, connector):
         data = {
             'driver_volume_type': 'vitastor',
@@ -538,9 +697,13 @@ class VitastorDriver(driver.CloneableImageVD,
                     size = int(volume.size) * units.Gi
                     dest_name = utils.convert_str(volume.name)
                     # Find or create the base snapshot
-                    self._cli('create base snapshot', 'create', '--allow-existing', '1', base_vol.name+'@.clone_snap')
+                    snap_cfg = self._create_snapshot(base_vol.name, base_vol.name+'@.clone_snap', True)
                     # Then create a clone from it
-                    self._cli('create clone', 'create', dest_name, '--size', size, '--parent', base_vol.name+'@.clone_snap')
+                    new_cfg = self._create_image(dest_name, {
+                        'size': size,
+                        'parent_id': snap_cfg['parent_id'],
+                        'parent_pool_id': snap_cfg['parent_pool_id'],
+                    })
                     return ({}, True)
         return ({}, False)
 
@@ -607,8 +770,26 @@ class VitastorDriver(driver.CloneableImageVD,
     def extend_volume(self, volume, new_size):
         """Extend an existing volume."""
         vol_name = utils.convert_str(volume.name)
-        size = int(new_size) * units.Gi
-        self._cli('extend volume', 'modify', vol_name, '--resize', new_size)
+        while True:
+            vol = self._get_image(vol_name)
+            if not vol:
+                raise exception.VolumeBackendAPIException(data = 'Volume '+vol_name+' does not exist')
+            # change size
+            size = int(new_size) * units.Gi
+            if size == vol['cfg']['size']:
+                break
+            resp = self._etcd_txn({ 'compare': [ {
+                'target': 'MOD',
+                'mod_revision': vol['cfg_mod'],
+                'key': 'config/inode/'+str(vol['idx']['pool_id'])+'/'+str(vol['idx']['id']),
+            } ], 'success': [
+                { 'request_put': {
+                    'key': 'config/inode/'+str(vol['idx']['pool_id'])+'/'+str(vol['idx']['id']),
+                    'value': json.dumps({ **vol['cfg'], 'size': size }),
+                } },
+            ] })
+            if resp.get('succeeded'):
+                break
         LOG.debug(
             "Extend volume from %(old_size)s GB to %(new_size)s GB.",
             {'old_size': volume.size, 'new_size': new_size}
@@ -681,7 +862,28 @@ class VitastorDriver(driver.CloneableImageVD,
         """
         from_name = self._get_existing_name(existing_ref)
         to_name = utils.convert_str(volume.name)
-        self._cli('rename', 'modify', from_name, '--rename', to_name)
+        self._rename(from_name, to_name)
+
+    def _rename(self, from_name, to_name):
+        while True:
+            vol = self._get_image(from_name)
+            if not vol:
+                raise exception.VolumeBackendAPIException(data = 'Volume '+from_name+' does not exist')
+            to = self._get_image(to_name)
+            if to:
+                raise exception.VolumeBackendAPIException(data = 'Volume '+to_name+' already exists')
+            resp = self._etcd_txn({ 'compare': [
+                { 'target': 'MOD', 'mod_revision': vol['idx_mod'], 'key': 'index/image/'+vol['cfg']['name'] },
+                { 'target': 'MOD', 'mod_revision': vol['cfg_mod'], 'key': 'config/inode/'+str(vol['idx']['pool_id'])+'/'+str(vol['idx']['id']) },
+                { 'target': 'VERSION', 'version': 0, 'key': 'index/image/'+to_name },
+            ], 'success': [
+                { 'request_delete_range': { 'key': 'index/image/'+vol['cfg']['name'] } },
+                { 'request_put': { 'key': 'index/image/'+to_name, 'value': json.dumps(vol['idx']) } },
+                { 'request_put': { 'key': 'config/inode/'+str(vol['idx']['pool_id'])+'/'+str(vol['idx']['id']),
+                    'value': json.dumps({ **vol['cfg'], 'name': to_name }) } },
+            ] })
+            if resp.get('succeeded'):
+                break
 
     def unmanage(self, volume):
         pass
@@ -754,7 +956,7 @@ class VitastorDriver(driver.CloneableImageVD,
         snap_name = self._get_existing_name(existing_ref)
         from_name = vol_name+'@'+snap_name
         to_name = vol_name+'@'+utils.convert_str(snapshot.name)
-        self._cli('rename', 'modify', from_name, '--rename', to_name)
+        self._rename(from_name, to_name)
 
     def unmanage_snapshot(self, snapshot):
         """Removes the specified snapshot from Cinder management."""

patches/libvirt-12.0-vitastor.diff

@@ -1,637 +0,0 @@
-diff --git a/include/libvirt/libvirt-storage.h b/include/libvirt/libvirt-storage.h
-index aaad4a3da1..5f5daa8341 100644
---- a/include/libvirt/libvirt-storage.h
-+++ b/include/libvirt/libvirt-storage.h
-@@ -326,6 +326,7 @@ typedef enum {
-     VIR_CONNECT_LIST_STORAGE_POOLS_ZFS           = 1 << 17, /* (Since: 1.2.8) */
-     VIR_CONNECT_LIST_STORAGE_POOLS_VSTORAGE      = 1 << 18, /* (Since: 3.1.0) */
-     VIR_CONNECT_LIST_STORAGE_POOLS_ISCSI_DIRECT  = 1 << 19, /* (Since: 5.6.0) */
-+    VIR_CONNECT_LIST_STORAGE_POOLS_VITASTOR      = 1 << 20, /* (Since: 5.0.0) */
- } virConnectListAllStoragePoolsFlags;
- 
- int                     virConnectListAllStoragePools(virConnectPtr conn,
-diff --git a/src/conf/domain_conf.c b/src/conf/domain_conf.c
-index 9ca5c2450c..cc52f00c0c 100644
---- a/src/conf/domain_conf.c
-+++ b/src/conf/domain_conf.c
-@@ -7453,7 +7453,8 @@ virDomainDiskSourceNetworkParse(xmlNodePtr node,
-     src->configFile = virXPathString("string(./config/@file)", ctxt);
- 
-     if (src->protocol == VIR_STORAGE_NET_PROTOCOL_HTTP ||
--        src->protocol == VIR_STORAGE_NET_PROTOCOL_HTTPS)
-+        src->protocol == VIR_STORAGE_NET_PROTOCOL_HTTPS ||
-+        src->protocol == VIR_STORAGE_NET_PROTOCOL_VITASTOR)
-         src->query = virXMLPropString(node, "query");
- 
-     if (virDomainStorageNetworkParseHosts(node, ctxt, &src->hosts, &src->nhosts) < 0)
-@@ -32187,6 +32188,7 @@ virDomainStorageSourceTranslateSourcePool(virStorageSource *src,
- 
-     case VIR_STORAGE_POOL_MPATH:
-     case VIR_STORAGE_POOL_RBD:
-+    case VIR_STORAGE_POOL_VITASTOR:
-     case VIR_STORAGE_POOL_SHEEPDOG:
-     case VIR_STORAGE_POOL_GLUSTER:
-     case VIR_STORAGE_POOL_LAST:
-diff --git a/src/conf/domain_validate.c b/src/conf/domain_validate.c
-index 7346a61731..83e94d762e 100644
---- a/src/conf/domain_validate.c
-+++ b/src/conf/domain_validate.c
-@@ -520,6 +520,7 @@ virDomainDiskDefValidateSourceChainOne(const virStorageSource *src)
-         case VIR_STORAGE_NET_PROTOCOL_RBD:
-             break;
- 
-+        case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
-         case VIR_STORAGE_NET_PROTOCOL_NBD:
-         case VIR_STORAGE_NET_PROTOCOL_SHEEPDOG:
-         case VIR_STORAGE_NET_PROTOCOL_GLUSTER:
-@@ -592,7 +593,7 @@ virDomainDiskDefValidateSourceChainOne(const virStorageSource *src)
-         }
-     }
- 
--    /* internal snapshots and config files are currently supported only with rbd: */
-+    /* internal snapshots are currently supported only with rbd: */
-     if (virStorageSourceGetActualType(src) != VIR_STORAGE_TYPE_NETWORK &&
-         src->protocol != VIR_STORAGE_NET_PROTOCOL_RBD) {
-         if (src->snapshot) {
-@@ -600,10 +601,14 @@ virDomainDiskDefValidateSourceChainOne(const virStorageSource *src)
-                            _("<snapshot> element is currently supported only with 'rbd' disks"));
-             return -1;
-         }
--
-+    }
-+    /* config files are currently supported only with rbd and vitastor: */
-+    if (virStorageSourceGetActualType(src) != VIR_STORAGE_TYPE_NETWORK &&
-+        src->protocol != VIR_STORAGE_NET_PROTOCOL_RBD &&
-+        src->protocol != VIR_STORAGE_NET_PROTOCOL_VITASTOR) {
-         if (src->configFile) {
-             virReportError(VIR_ERR_XML_ERROR, "%s",
--                           _("<config> element is currently supported only with 'rbd' disks"));
-+                           _("<config> element is currently supported only with 'rbd' and 'vitastor' disks"));
-             return -1;
-         }
-     }
-diff --git a/src/conf/schemas/domaincommon.rng b/src/conf/schemas/domaincommon.rng
-index 114dd3f96f..c71f9a3277 100644
---- a/src/conf/schemas/domaincommon.rng
-+++ b/src/conf/schemas/domaincommon.rng
-@@ -2093,6 +2093,35 @@
-     </element>
-   </define>
- 
-+  <define name="diskSourceNetworkProtocolVitastor">
-+    <element name="source">
-+      <interleave>
-+        <attribute name="protocol">
-+          <value>vitastor</value>
-+        </attribute>
-+        <ref name="diskSourceCommon"/>
-+        <optional>
-+          <attribute name="name"/>
-+        </optional>
-+        <optional>
-+          <attribute name="query"/>
-+        </optional>
-+        <zeroOrMore>
-+          <ref name="diskSourceNetworkHost"/>
-+        </zeroOrMore>
-+        <optional>
-+          <element name="config">
-+            <attribute name="file">
-+              <ref name="absFilePath"/>
-+            </attribute>
-+            <empty/>
-+          </element>
-+        </optional>
-+        <empty/>
-+      </interleave>
-+    </element>
-+  </define>
-+
-   <define name="diskSourceNetworkProtocolISCSI">
-     <element name="source">
-       <attribute name="protocol">
-@@ -2443,6 +2472,7 @@
-       <ref name="diskSourceNetworkProtocolSimple"/>
-       <ref name="diskSourceNetworkProtocolVxHS"/>
-       <ref name="diskSourceNetworkProtocolNFS"/>
-+      <ref name="diskSourceNetworkProtocolVitastor"/>
-     </choice>
-   </define>
- 
-diff --git a/src/conf/storage_conf.c b/src/conf/storage_conf.c
-index 1dc9365bf2..a8a736be81 100644
---- a/src/conf/storage_conf.c
-+++ b/src/conf/storage_conf.c
-@@ -56,7 +56,7 @@ VIR_ENUM_IMPL(virStoragePool,
-               "logical", "disk", "iscsi",
-               "iscsi-direct", "scsi", "mpath",
-               "rbd", "sheepdog", "gluster",
--              "zfs", "vstorage",
-+              "zfs", "vstorage", "vitastor",
- );
- 
- VIR_ENUM_IMPL(virStoragePoolFormatFileSystem,
-@@ -242,6 +242,18 @@ static virStoragePoolTypeInfo poolTypeInfo[] = {
-           .formatToString = virStorageFileFormatTypeToString,
-       }
-     },
-+    {.poolType = VIR_STORAGE_POOL_VITASTOR,
-+     .poolOptions = {
-+         .flags = (VIR_STORAGE_POOL_SOURCE_HOST |
-+                   VIR_STORAGE_POOL_SOURCE_NETWORK |
-+                   VIR_STORAGE_POOL_SOURCE_NAME),
-+      },
-+      .volOptions = {
-+          .defaultFormat = VIR_STORAGE_FILE_RAW,
-+          .formatFromString = virStorageVolumeFormatFromString,
-+          .formatToString = virStorageFileFormatTypeToString,
-+      }
-+    },
-     {.poolType = VIR_STORAGE_POOL_SHEEPDOG,
-      .poolOptions = {
-          .flags = (VIR_STORAGE_POOL_SOURCE_HOST |
-@@ -538,6 +550,11 @@ virStoragePoolDefParseSource(xmlXPathContextPtr ctxt,
-                        _("element 'name' is mandatory for RBD pool"));
-         return -1;
-     }
-+    if (pool_type == VIR_STORAGE_POOL_VITASTOR && source->name == NULL) {
-+        virReportError(VIR_ERR_XML_ERROR, "%s",
-+                       _("element 'name' is mandatory for Vitastor pool"));
-+        return -1;
-+    }
- 
-     if (options->formatFromString) {
-         g_autofree char *format = NULL;
-@@ -1127,6 +1144,7 @@ virStoragePoolDefFormatBuf(virBuffer *buf,
-     /* RBD, Sheepdog, Gluster and Iscsi-direct devices are not local block devs nor
-      * files, so they don't have a target */
-     if (def->type != VIR_STORAGE_POOL_RBD &&
-+        def->type != VIR_STORAGE_POOL_VITASTOR &&
-         def->type != VIR_STORAGE_POOL_SHEEPDOG &&
-         def->type != VIR_STORAGE_POOL_GLUSTER &&
-         def->type != VIR_STORAGE_POOL_ISCSI_DIRECT) {
-diff --git a/src/conf/storage_conf.h b/src/conf/storage_conf.h
-index fc67957cfe..720c07ef74 100644
---- a/src/conf/storage_conf.h
-+++ b/src/conf/storage_conf.h
-@@ -103,6 +103,7 @@ typedef enum {
-     VIR_STORAGE_POOL_GLUSTER,  /* Gluster device */
-     VIR_STORAGE_POOL_ZFS,      /* ZFS */
-     VIR_STORAGE_POOL_VSTORAGE, /* Virtuozzo Storage */
-+    VIR_STORAGE_POOL_VITASTOR, /* Vitastor */
- 
-     VIR_STORAGE_POOL_LAST,
- } virStoragePoolType;
-@@ -454,6 +455,7 @@ VIR_ENUM_DECL(virStoragePartedFs);
-                  VIR_CONNECT_LIST_STORAGE_POOLS_SCSI     | \
-                  VIR_CONNECT_LIST_STORAGE_POOLS_MPATH    | \
-                  VIR_CONNECT_LIST_STORAGE_POOLS_RBD      | \
-+                 VIR_CONNECT_LIST_STORAGE_POOLS_VITASTOR | \
-                  VIR_CONNECT_LIST_STORAGE_POOLS_SHEEPDOG | \
-                  VIR_CONNECT_LIST_STORAGE_POOLS_GLUSTER  | \
-                  VIR_CONNECT_LIST_STORAGE_POOLS_ZFS      | \
-diff --git a/src/conf/storage_source_conf.c b/src/conf/storage_source_conf.c
-index d7b9bdfecb..38aefd0dd4 100644
---- a/src/conf/storage_source_conf.c
-+++ b/src/conf/storage_source_conf.c
-@@ -90,6 +90,7 @@ VIR_ENUM_IMPL(virStorageNetProtocol,
-               "ssh",
-               "vxhs",
-               "nfs",
-+              "vitastor",
- );
- 
- 
-@@ -1317,6 +1318,7 @@ virStorageSourceNetworkDefaultPort(virStorageNetProtocol protocol)
-         case VIR_STORAGE_NET_PROTOCOL_GLUSTER:
-             return 24007;
- 
-+        case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
-         case VIR_STORAGE_NET_PROTOCOL_RBD:
-             /* we don't provide a default for RBD */
-             return 0;
-diff --git a/src/conf/storage_source_conf.h b/src/conf/storage_source_conf.h
-index 22c35d420d..f1e32ea83d 100644
---- a/src/conf/storage_source_conf.h
-+++ b/src/conf/storage_source_conf.h
-@@ -131,6 +131,7 @@ typedef enum {
-     VIR_STORAGE_NET_PROTOCOL_SSH,
-     VIR_STORAGE_NET_PROTOCOL_VXHS,
-     VIR_STORAGE_NET_PROTOCOL_NFS,
-+    VIR_STORAGE_NET_PROTOCOL_VITASTOR,
- 
-     VIR_STORAGE_NET_PROTOCOL_LAST
- } virStorageNetProtocol;
-diff --git a/src/conf/virstorageobj.c b/src/conf/virstorageobj.c
-index 59fa5da372..4739167f5f 100644
---- a/src/conf/virstorageobj.c
-+++ b/src/conf/virstorageobj.c
-@@ -1438,6 +1438,7 @@ virStoragePoolObjSourceFindDuplicateCb(const void *payload,
-             return 1;
-         break;
- 
-+    case VIR_STORAGE_POOL_VITASTOR:
-     case VIR_STORAGE_POOL_ISCSI_DIRECT:
-     case VIR_STORAGE_POOL_RBD:
-     case VIR_STORAGE_POOL_LAST:
-@@ -1921,6 +1922,8 @@ virStoragePoolObjMatch(virStoragePoolObj *obj,
-                (obj->def->type == VIR_STORAGE_POOL_MPATH))   ||
-               (MATCH(VIR_CONNECT_LIST_STORAGE_POOLS_RBD) &&
-                (obj->def->type == VIR_STORAGE_POOL_RBD))     ||
-+              (MATCH(VIR_CONNECT_LIST_STORAGE_POOLS_VITASTOR) &&
-+               (obj->def->type == VIR_STORAGE_POOL_VITASTOR)) ||
-               (MATCH(VIR_CONNECT_LIST_STORAGE_POOLS_SHEEPDOG) &&
-                (obj->def->type == VIR_STORAGE_POOL_SHEEPDOG)) ||
-               (MATCH(VIR_CONNECT_LIST_STORAGE_POOLS_GLUSTER) &&
-diff --git a/src/libvirt-storage.c b/src/libvirt-storage.c
-index db7660aac4..561df34709 100644
---- a/src/libvirt-storage.c
-+++ b/src/libvirt-storage.c
-@@ -94,6 +94,7 @@ virStoragePoolGetConnect(virStoragePoolPtr pool)
-  * VIR_CONNECT_LIST_STORAGE_POOLS_SCSI
-  * VIR_CONNECT_LIST_STORAGE_POOLS_MPATH
-  * VIR_CONNECT_LIST_STORAGE_POOLS_RBD
-+ * VIR_CONNECT_LIST_STORAGE_POOLS_VITASTOR
-  * VIR_CONNECT_LIST_STORAGE_POOLS_SHEEPDOG
-  * VIR_CONNECT_LIST_STORAGE_POOLS_GLUSTER
-  * VIR_CONNECT_LIST_STORAGE_POOLS_ZFS
-diff --git a/src/libxl/libxl_conf.c b/src/libxl/libxl_conf.c
-index 2b988157fa..9d0eb47b25 100644
---- a/src/libxl/libxl_conf.c
-+++ b/src/libxl/libxl_conf.c
-@@ -1069,6 +1069,7 @@ libxlMakeNetworkDiskSrcStr(virStorageSource *src,
-     case VIR_STORAGE_NET_PROTOCOL_SSH:
-     case VIR_STORAGE_NET_PROTOCOL_VXHS:
-     case VIR_STORAGE_NET_PROTOCOL_NFS:
-+    case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
-     case VIR_STORAGE_NET_PROTOCOL_LAST:
-     case VIR_STORAGE_NET_PROTOCOL_NONE:
-         virReportError(VIR_ERR_NO_SUPPORT,
-diff --git a/src/libxl/xen_xl.c b/src/libxl/xen_xl.c
-index e72e7d7f44..8482c21805 100644
---- a/src/libxl/xen_xl.c
-+++ b/src/libxl/xen_xl.c
-@@ -1461,6 +1461,7 @@ xenFormatXLDiskSrcNet(virStorageSource *src)
-     case VIR_STORAGE_NET_PROTOCOL_SSH:
-     case VIR_STORAGE_NET_PROTOCOL_VXHS:
-     case VIR_STORAGE_NET_PROTOCOL_NFS:
-+    case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
-     case VIR_STORAGE_NET_PROTOCOL_LAST:
-     case VIR_STORAGE_NET_PROTOCOL_NONE:
-         virReportError(VIR_ERR_NO_SUPPORT,
-diff --git a/src/qemu/qemu_block.c b/src/qemu/qemu_block.c
-index 9b43279797..459d8e8a65 100644
---- a/src/qemu/qemu_block.c
-+++ b/src/qemu/qemu_block.c
-@@ -743,6 +743,38 @@ qemuBlockStorageSourceGetRBDProps(virStorageSource *src,
- }
- 
- 
-+static virJSONValue *
-+qemuBlockStorageSourceGetVitastorProps(virStorageSource *src)
-+{
-+    virJSONValue *ret = NULL;
-+    virStorageNetHostDef *host;
-+    size_t i;
-+    g_auto(virBuffer) buf = VIR_BUFFER_INITIALIZER;
-+    g_autofree char *etcd = NULL;
-+
-+    for (i = 0; i < src->nhosts; i++) {
-+        host = src->hosts + i;
-+        if ((virStorageNetHostTransport)host->transport != VIR_STORAGE_NET_HOST_TRANS_TCP) {
-+            return NULL;
-+        }
-+        virBufferAsprintf(&buf, i > 0 ? ",%s:%u" : "%s:%u", host->name, host->port);
-+    }
-+    if (src->nhosts > 0) {
-+        etcd = virBufferContentAndReset(&buf);
-+    }
-+
-+    if (virJSONValueObjectAdd(&ret,
-+                              "S:etcd-host", etcd,
-+                              "S:etcd-prefix", src->query,
-+                              "S:config-path", src->configFile,
-+                              "s:image", src->path,
-+                              NULL) < 0)
-+        return NULL;
-+
-+    return ret;
-+}
-+
-+
- static virJSONValue *
- qemuBlockStorageSourceGetSshProps(virStorageSource *src)
- {
-@@ -1094,6 +1126,12 @@ qemuBlockStorageSourceGetBackendProps(virStorageSource *src,
-                 return NULL;
-             break;
- 
-+        case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
-+            driver = "vitastor";
-+            if (!(fileprops = qemuBlockStorageSourceGetVitastorProps(src)))
-+                return NULL;
-+            break;
-+
-         case VIR_STORAGE_NET_PROTOCOL_SSH:
-             driver = "ssh";
-             if (!(fileprops = qemuBlockStorageSourceGetSshProps(src)))
-@@ -1997,6 +2035,7 @@ qemuBlockGetBackingStoreString(virStorageSource *src,
- 
-             case VIR_STORAGE_NET_PROTOCOL_SHEEPDOG:
-             case VIR_STORAGE_NET_PROTOCOL_RBD:
-+            case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
-             case VIR_STORAGE_NET_PROTOCOL_VXHS:
-             case VIR_STORAGE_NET_PROTOCOL_NFS:
-             case VIR_STORAGE_NET_PROTOCOL_SSH:
-@@ -2377,6 +2416,12 @@ qemuBlockStorageSourceCreateGetStorageProps(virStorageSource *src,
-                 return -1;
-             break;
- 
-+        case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
-+            driver = "vitastor";
-+            if (!(location = qemuBlockStorageSourceGetVitastorProps(src)))
-+                return -1;
-+            break;
-+
-         case VIR_STORAGE_NET_PROTOCOL_SSH:
-             if (srcPriv->nbdkitProcess) {
-                 /* disk creation not yet supported with nbdkit, and even if it
-diff --git a/src/qemu/qemu_domain.c b/src/qemu/qemu_domain.c
-index ac56fc7cb4..9e407b4aab 100644
---- a/src/qemu/qemu_domain.c
-+++ b/src/qemu/qemu_domain.c
-@@ -4677,7 +4677,8 @@ qemuDomainValidateStorageSource(virStorageSource *src,
-     if (src->query &&
-         (actualType != VIR_STORAGE_TYPE_NETWORK ||
-          (src->protocol != VIR_STORAGE_NET_PROTOCOL_HTTPS &&
--          src->protocol != VIR_STORAGE_NET_PROTOCOL_HTTP))) {
-+          src->protocol != VIR_STORAGE_NET_PROTOCOL_HTTP &&
-+          src->protocol != VIR_STORAGE_NET_PROTOCOL_VITASTOR))) {
-         virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
-                        _("query is supported only with HTTP(S) protocols"));
-         return -1;
-@@ -9103,6 +9104,7 @@ qemuDomainPrepareStorageSourceTLS(virStorageSource *src,
-         break;
- 
-     case VIR_STORAGE_NET_PROTOCOL_RBD:
-+    case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
-     case VIR_STORAGE_NET_PROTOCOL_SHEEPDOG:
-     case VIR_STORAGE_NET_PROTOCOL_GLUSTER:
-     case VIR_STORAGE_NET_PROTOCOL_ISCSI:
-diff --git a/src/qemu/qemu_snapshot.c b/src/qemu/qemu_snapshot.c
-index e738afffc3..37d64f469b 100644
---- a/src/qemu/qemu_snapshot.c
-+++ b/src/qemu/qemu_snapshot.c
-@@ -665,6 +665,7 @@ qemuSnapshotPrepareDiskExternalInactive(virDomainSnapshotDiskDef *snapdisk,
-         case VIR_STORAGE_NET_PROTOCOL_NONE:
-         case VIR_STORAGE_NET_PROTOCOL_NBD:
-         case VIR_STORAGE_NET_PROTOCOL_RBD:
-+        case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
-         case VIR_STORAGE_NET_PROTOCOL_SHEEPDOG:
-         case VIR_STORAGE_NET_PROTOCOL_GLUSTER:
-         case VIR_STORAGE_NET_PROTOCOL_ISCSI:
-@@ -893,6 +894,7 @@ qemuSnapshotPrepareDiskInternal(virDomainDiskDef *disk,
-         case VIR_STORAGE_NET_PROTOCOL_NONE:
-         case VIR_STORAGE_NET_PROTOCOL_NBD:
-         case VIR_STORAGE_NET_PROTOCOL_RBD:
-+        case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
-         case VIR_STORAGE_NET_PROTOCOL_SHEEPDOG:
-         case VIR_STORAGE_NET_PROTOCOL_GLUSTER:
-         case VIR_STORAGE_NET_PROTOCOL_ISCSI:
-diff --git a/src/storage/storage_driver.c b/src/storage/storage_driver.c
-index e19e032427..59f91f4710 100644
---- a/src/storage/storage_driver.c
-+++ b/src/storage/storage_driver.c
-@@ -1626,6 +1626,7 @@ storageVolLookupByPathCallback(virStoragePoolObj *obj,
- 
-         case VIR_STORAGE_POOL_GLUSTER:
-         case VIR_STORAGE_POOL_RBD:
-+        case VIR_STORAGE_POOL_VITASTOR:
-         case VIR_STORAGE_POOL_SHEEPDOG:
-         case VIR_STORAGE_POOL_ZFS:
-         case VIR_STORAGE_POOL_LAST:
-diff --git a/src/storage_file/storage_source_backingstore.c b/src/storage_file/storage_source_backingstore.c
-index 821378883c..2211f6891b 100644
---- a/src/storage_file/storage_source_backingstore.c
-+++ b/src/storage_file/storage_source_backingstore.c
-@@ -264,6 +264,75 @@ virStorageSourceParseRBDColonString(const char *rbdstr,
- }
- 
- 
-+static int
-+virStorageSourceParseVitastorColonString(const char *colonstr,
-+                                         virStorageSource *src)
-+{
-+    char *p, *e, *next;
-+    g_autofree char *options = NULL;
-+
-+    /* optionally skip the "vitastor:" prefix if provided */
-+    if (STRPREFIX(colonstr, "vitastor:"))
-+        colonstr += strlen("vitastor:");
-+
-+    options = g_strdup(colonstr);
-+
-+    p = options;
-+    while (*p) {
-+        /* find : delimiter or end of string */
-+        for (e = p; *e && *e != ':'; ++e) {
-+            if (*e == '\\') {
-+                e++;
-+                if (*e == '\0')
-+                    break;
-+            }
-+        }
-+        if (*e == '\0') {
-+            next = e;    /* last kv pair */
-+        } else {
-+            next = e + 1;
-+            *e = '\0';
-+        }
-+
-+        if (STRPREFIX(p, "image=")) {
-+            src->path = g_strdup(p + strlen("image="));
-+        } else if (STRPREFIX(p, "etcd-prefix=")) {
-+            src->query = g_strdup(p + strlen("etcd-prefix="));
-+        } else if (STRPREFIX(p, "config-path=")) {
-+            src->configFile = g_strdup(p + strlen("config-path="));
-+        } else if (STRPREFIX(p, "etcd-host=")) {
-+            char *h, *sep;
-+
-+            h = p + strlen("etcd-host=");
-+            while (h < e) {
-+                for (sep = h; sep < e; ++sep) {
-+                    if (*sep == '\\' && (sep[1] == ',' ||
-+                                         sep[1] == ';' ||
-+                                         sep[1] == ' ')) {
-+                        *sep = '\0';
-+                        sep += 2;
-+                        break;
-+                    }
-+                }
-+
-+                if (virStorageSourceRBDAddHost(src, h) < 0)
-+                    return -1;
-+
-+                h = sep;
-+            }
-+        }
-+
-+        p = next;
-+    }
-+
-+    if (!src->path) {
-+        return -1;
-+    }
-+
-+    return 0;
-+}
-+
-+
- static int
- virStorageSourceParseNBDColonString(const char *nbdstr,
-                                     virStorageSource *src)
-@@ -379,6 +448,11 @@ virStorageSourceParseBackingColon(virStorageSource *src,
-             return -1;
-         break;
- 
-+    case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
-+        if (virStorageSourceParseVitastorColonString(path, src) < 0)
-+            return -1;
-+        break;
-+
-     case VIR_STORAGE_NET_PROTOCOL_SHEEPDOG:
-     case VIR_STORAGE_NET_PROTOCOL_LAST:
-     case VIR_STORAGE_NET_PROTOCOL_NONE:
-@@ -953,6 +1027,54 @@ virStorageSourceParseBackingJSONRBD(virStorageSource *src,
-     return 0;
- }
- 
-+static int
-+virStorageSourceParseBackingJSONVitastor(virStorageSource *src,
-+                                         virJSONValue *json,
-+                                         const char *jsonstr G_GNUC_UNUSED,
-+                                         int opaque G_GNUC_UNUSED)
-+{
-+    const char *filename;
-+    const char *image = virJSONValueObjectGetString(json, "image");
-+    const char *conf = virJSONValueObjectGetString(json, "config-path");
-+    const char *etcd_prefix = virJSONValueObjectGetString(json, "etcd-prefix");
-+    virJSONValue *servers = virJSONValueObjectGetArray(json, "server");
-+    size_t nservers;
-+    size_t i;
-+
-+    src->type = VIR_STORAGE_TYPE_NETWORK;
-+    src->protocol = VIR_STORAGE_NET_PROTOCOL_VITASTOR;
-+
-+    /* legacy syntax passed via 'filename' option */
-+    if ((filename = virJSONValueObjectGetString(json, "filename")))
-+        return virStorageSourceParseVitastorColonString(filename, src);
-+
-+    if (!image) {
-+        virReportError(VIR_ERR_INVALID_ARG, "%s",
-+                       _("missing image name in Vitastor backing volume "
-+                         "JSON specification"));
-+        return -1;
-+    }
-+
-+    src->path = g_strdup(image);
-+    src->configFile = g_strdup(conf);
-+    src->query = g_strdup(etcd_prefix);
-+
-+    if (servers) {
-+        nservers = virJSONValueArraySize(servers);
-+
-+        src->hosts = g_new0(virStorageNetHostDef, nservers);
-+        src->nhosts = nservers;
-+
-+        for (i = 0; i < nservers; i++) {
-+            if (virStorageSourceParseBackingJSONInetSocketAddress(src->hosts + i,
-+                                                                  virJSONValueArrayGet(servers, i)) < 0)
-+                return -1;
-+        }
-+    }
-+
-+    return 0;
-+}
-+
- static int
- virStorageSourceParseBackingJSONRaw(virStorageSource *src,
-                                     virJSONValue *json,
-@@ -1130,6 +1252,7 @@ static const struct virStorageSourceJSONDriverParser jsonParsers[] = {
-     {"sheepdog", false, virStorageSourceParseBackingJSONSheepdog, 0},
-     {"ssh", false, virStorageSourceParseBackingJSONSSH, 0},
-     {"rbd", false, virStorageSourceParseBackingJSONRBD, 0},
-+    {"vitastor", false, virStorageSourceParseBackingJSONVitastor, 0},
-     {"raw", true, virStorageSourceParseBackingJSONRaw, 0},
-     {"nfs", false, virStorageSourceParseBackingJSONNFS, 0},
-     {"vxhs", false, virStorageSourceParseBackingJSONVxHS, 0},
-diff --git a/src/test/test_driver.c b/src/test/test_driver.c
-index 1165689de7..bba846351c 100644
---- a/src/test/test_driver.c
-+++ b/src/test/test_driver.c
-@@ -7345,6 +7345,7 @@ testStorageVolumeTypeForPool(int pooltype)
-     case VIR_STORAGE_POOL_ISCSI_DIRECT:
-     case VIR_STORAGE_POOL_GLUSTER:
-     case VIR_STORAGE_POOL_RBD:
-+    case VIR_STORAGE_POOL_VITASTOR:
-         return VIR_STORAGE_VOL_NETWORK;
-     case VIR_STORAGE_POOL_LOGICAL:
-     case VIR_STORAGE_POOL_DISK:
-diff --git a/tests/storagepoolcapsschemadata/poolcaps-fs.xml b/tests/storagepoolcapsschemadata/poolcaps-fs.xml
-index eee75af746..8bd0a57bdd 100644
---- a/tests/storagepoolcapsschemadata/poolcaps-fs.xml
-+++ b/tests/storagepoolcapsschemadata/poolcaps-fs.xml
-@@ -204,4 +204,11 @@
-       </enum>
-     </volOptions>
-   </pool>
-+  <pool type='vitastor' supported='no'>
-+    <volOptions>
-+      <defaultFormat type='raw'/>
-+      <enum name='targetFormatType'>
-+      </enum>
-+    </volOptions>
-+  </pool>
- </storagepoolCapabilities>
-diff --git a/tests/storagepoolcapsschemadata/poolcaps-full.xml b/tests/storagepoolcapsschemadata/poolcaps-full.xml
-index 805950a937..852df0de16 100644
---- a/tests/storagepoolcapsschemadata/poolcaps-full.xml
-+++ b/tests/storagepoolcapsschemadata/poolcaps-full.xml
-@@ -204,4 +204,11 @@
-       </enum>
-     </volOptions>
-   </pool>
-+  <pool type='vitastor' supported='yes'>
-+    <volOptions>
-+      <defaultFormat type='raw'/>
-+      <enum name='targetFormatType'>
-+      </enum>
-+    </volOptions>
-+  </pool>
- </storagepoolCapabilities>
-diff --git a/tests/storagepoolxml2argvtest.c b/tests/storagepoolxml2argvtest.c
-index d5c2531ab8..b19308ac38 100644
---- a/tests/storagepoolxml2argvtest.c
-+++ b/tests/storagepoolxml2argvtest.c
-@@ -57,6 +57,7 @@ testCompareXMLToArgvFiles(bool shouldFail,
-     case VIR_STORAGE_POOL_GLUSTER:
-     case VIR_STORAGE_POOL_ZFS:
-     case VIR_STORAGE_POOL_VSTORAGE:
-+    case VIR_STORAGE_POOL_VITASTOR:
-     case VIR_STORAGE_POOL_LAST:
-     default:
-         VIR_TEST_DEBUG("pool type '%s' has no xml2argv test", defTypeStr);
-diff --git a/tools/virsh-pool.c b/tools/virsh-pool.c
-index 2010ef1356..072e2ff9e8 100644
---- a/tools/virsh-pool.c
-+++ b/tools/virsh-pool.c
-@@ -1187,6 +1187,9 @@ cmdPoolList(vshControl *ctl, const vshCmd *cmd G_GNUC_UNUSED)
-             case VIR_STORAGE_POOL_VSTORAGE:
-                 flags |= VIR_CONNECT_LIST_STORAGE_POOLS_VSTORAGE;
-                 break;
-+            case VIR_STORAGE_POOL_VITASTOR:
-+                flags |= VIR_CONNECT_LIST_STORAGE_POOLS_VITASTOR;
-+                break;
-             case VIR_STORAGE_POOL_LAST:
-                 break;
-             }

patches/linux-fix-atomic-write-checks.diff

@@ -1,39 +0,0 @@
-From 98d3f68a40130c438854f61db6025f9e9b099cb6 Mon Sep 17 00:00:00 2001
-From: Vitaliy Filippov <vitalifster@gmail.com>
-Date: Sat, 20 Dec 2025 14:44:35 +0300
-Subject: [PATCH] Do not require atomic writes to be power of 2 sized and
- aligned on length boundary
-
-It contradicts NVMe specification where alignment is only required when atomic
-write boundary (NABSPF/NABO) is set and highly limits usage of NVMe atomic writes
-
-Signed-off-by: Vitaliy Filippov <vitalifster@gmail.com>
----
- fs/read_write.c | 8 --------
- 1 file changed, 8 deletions(-)
-
-diff --git a/fs/read_write.c b/fs/read_write.c
-index 833bae068770..5467d710108d 100644
---- a/fs/read_write.c
-+++ b/fs/read_write.c
-@@ -1802,17 +1802,9 @@ int generic_file_rw_checks(struct file *file_in, struct file *file_out)
- 
- int generic_atomic_write_valid(struct kiocb *iocb, struct iov_iter *iter)
- {
--	size_t len = iov_iter_count(iter);
--
- 	if (!iter_is_ubuf(iter))
- 		return -EINVAL;
- 
--	if (!is_power_of_2(len))
--		return -EINVAL;
--
--	if (!IS_ALIGNED(iocb->ki_pos, len))
--		return -EINVAL;
--
- 	if (!(iocb->ki_flags & IOCB_DIRECT))
- 		return -EOPNOTSUPP;
- 
--- 
-2.51.0
-

patches/qemu-10.1-vitastor.patch

@@ -1,172 +1,29 @@
-diff --git a/block/meson.build b/block/meson.build
-index 34b1b2a306..24ca0f1e52 100644
---- a/block/meson.build
-+++ b/block/meson.build
-@@ -114,6 +114,7 @@ foreach m : [
-   [libnfs, 'nfs', files('nfs.c')],
-   [libssh, 'ssh', files('ssh.c')],
-   [rbd, 'rbd', files('rbd.c')],
-+  [vitastor, 'vitastor', files('vitastor.c')],
- ]
-   if m[0].found()
-     module_ss = ss.source_set()
-diff --git a/meson.build b/meson.build
-index 50c774a195..e5c7a3a4b1 100644
---- a/meson.build
-+++ b/meson.build
-@@ -1652,6 +1652,26 @@ if not get_option('rbd').auto() or have_block
-   endif
- endif
+diff --git a/src/client/qemu_driver.c b/src/client/qemu_driver.c
+index d8356dab..5f4cd50d 100644
+--- a/src/client/qemu_driver.c
++++ b/src/client/qemu_driver.c
+@@ -974,14 +974,21 @@ static void vitastor_co_read_bitmap_cb(void *opaque, long retval, uint8_t *bitma
+ #endif
+ }
  
-+vitastor = not_found
-+if not get_option('vitastor').auto() or have_block
-+  libvitastor_client = cc.find_library('vitastor_client', has_headers: ['vitastor_c.h'],
-+    required: get_option('vitastor'))
-+  if libvitastor_client.found()
-+    if cc.links('''
-+      #include <vitastor_c.h>
-+      int main(void) {
-+        vitastor_c_create_qemu(0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0);
-+        return 0;
-+      }''', dependencies: libvitastor_client)
-+      vitastor = declare_dependency(dependencies: libvitastor_client)
-+    elif get_option('vitastor').enabled()
-+      error('could not link libvitastor_client')
-+    else
-+      warning('could not link libvitastor_client, disabling')
-+    endif
-+  endif
-+endif
-+
- glusterfs = not_found
- glusterfs_ftruncate_has_stat = false
- glusterfs_iocb_has_stat = false
-@@ -2547,6 +2567,7 @@ endif
- config_host_data.set('CONFIG_OPENGL', opengl.found())
- config_host_data.set('CONFIG_PLUGIN', get_option('plugins'))
- config_host_data.set('CONFIG_RBD', rbd.found())
-+config_host_data.set('CONFIG_VITASTOR', vitastor.found())
- config_host_data.set('CONFIG_RDMA', rdma.found())
- config_host_data.set('CONFIG_RELOCATABLE', get_option('relocatable'))
- config_host_data.set('CONFIG_SAFESTACK', get_option('safe_stack'))
-@@ -4972,6 +4993,7 @@ summary_info += {'fdt support':       fdt_opt == 'internal' ? 'internal' : fdt}
- summary_info += {'libcap-ng support': libcap_ng}
- summary_info += {'bpf support':       libbpf}
- summary_info += {'rbd support':       rbd}
-+summary_info += {'vitastor support':  vitastor}
- summary_info += {'smartcard support': cacard}
- summary_info += {'U2F support':       u2f}
- summary_info += {'libusb':            libusb}
-diff --git a/meson_options.txt b/meson_options.txt
-index fff1521e58..f0844c0e00 100644
---- a/meson_options.txt
-+++ b/meson_options.txt
-@@ -202,6 +202,8 @@ option('pvg', type: 'feature', value: 'auto',
-        description: 'macOS paravirtualized graphics support')
- option('rbd', type : 'feature', value : 'auto',
-        description: 'Ceph block device driver')
-+option('vitastor', type : 'feature', value : 'auto',
-+       description: 'Vitastor block device driver')
- option('opengl', type : 'feature', value : 'auto',
-        description: 'OpenGL support')
- option('rdma', type : 'feature', value : 'auto',
-diff --git a/qapi/block-core.json b/qapi/block-core.json
-index dc6eb4ae23..d043f4340e 100644
---- a/qapi/block-core.json
-+++ b/qapi/block-core.json
-@@ -3280,7 +3280,7 @@
-             'parallels', 'preallocate', 'qcow', 'qcow2', 'qed', 'quorum',
-             'raw', 'rbd',
-             { 'name': 'replication', 'if': 'CONFIG_REPLICATION' },
--            'ssh', 'throttle', 'vdi', 'vhdx',
-+            'ssh', 'throttle', 'vdi', 'vhdx', 'vitastor',
-             { 'name': 'virtio-blk-vfio-pci', 'if': 'CONFIG_BLKIO' },
-             { 'name': 'virtio-blk-vhost-user', 'if': 'CONFIG_BLKIO' },
-             { 'name': 'virtio-blk-vhost-vdpa', 'if': 'CONFIG_BLKIO' },
-@@ -4363,6 +4363,28 @@
-             '*key-secret': 'str',
-             '*server': ['InetSocketAddressBase'] } }
- 
-+##
-+# @BlockdevOptionsVitastor:
-+#
-+# Driver specific block device options for vitastor
-+#
-+# @image:       Image name
-+# @inode:       Inode number
-+# @pool:        Pool ID
-+# @size:        Desired image size in bytes
-+# @config-path: Path to Vitastor configuration
-+# @etcd-host:   etcd connection address(es)
-+# @etcd-prefix: etcd key/value prefix
-+##
-+{ 'struct': 'BlockdevOptionsVitastor',
-+  'data': { '*inode': 'uint64',
-+            '*pool': 'uint64',
-+            '*size': 'uint64',
-+            '*image': 'str',
-+            '*config-path': 'str',
-+            '*etcd-host': 'str',
-+            '*etcd-prefix': 'str' } }
-+
- ##
- # @ReplicationMode:
- #
-@@ -4831,6 +4853,7 @@
-       'throttle':   'BlockdevOptionsThrottle',
-       'vdi':        'BlockdevOptionsGenericFormat',
-       'vhdx':       'BlockdevOptionsGenericFormat',
-+      'vitastor':   'BlockdevOptionsVitastor',
-       'virtio-blk-vfio-pci':
-                     { 'type': 'BlockdevOptionsVirtioBlkVfioPci',
-                       'if': 'CONFIG_BLKIO' },
-@@ -5304,6 +5327,20 @@
-             '*cluster-size' :   'size',
-             '*encrypt' :        'RbdEncryptionCreateOptions' } }
- 
-+##
-+# @BlockdevCreateOptionsVitastor:
-+#
-+# Driver specific image creation options for Vitastor.
-+#
-+# @location: Where to store the new image file.  This location cannot
-+#     point to a snapshot.
-+#
-+# @size: Size of the virtual disk in bytes
-+##
-+{ 'struct': 'BlockdevCreateOptionsVitastor',
-+  'data': { 'location':         'BlockdevOptionsVitastor',
-+            'size':             'size' } }
-+
- ##
- # @BlockdevVmdkSubformat:
- #
-@@ -5526,6 +5563,7 @@
-       'ssh':            'BlockdevCreateOptionsSsh',
-       'vdi':            'BlockdevCreateOptionsVdi',
-       'vhdx':           'BlockdevCreateOptionsVhdx',
-+      'vitastor':       'BlockdevCreateOptionsVitastor',
-       'vmdk':           'BlockdevCreateOptionsVmdk',
-       'vpc':            'BlockdevCreateOptionsVpc'
-   } }
-diff --git a/scripts/meson-buildoptions.sh b/scripts/meson-buildoptions.sh
-index 0ebe6bc52a..2c37ad3892 100644
---- a/scripts/meson-buildoptions.sh
-+++ b/scripts/meson-buildoptions.sh
-@@ -175,6 +175,7 @@ meson_options_help() {
-   printf "%s\n" '  qga-vss         build QGA VSS support (broken with MinGW)'
-   printf "%s\n" '  qpl             Query Processing Library support'
-   printf "%s\n" '  rbd             Ceph block device driver'
-+  printf "%s\n" '  vitastor        Vitastor block device driver'
-   printf "%s\n" '  rdma            Enable RDMA-based migration'
-   printf "%s\n" '  replication     replication support'
-   printf "%s\n" '  rust            Rust support'
-@@ -459,6 +460,8 @@ _meson_option_parse() {
-     --disable-qpl) printf "%s" -Dqpl=disabled ;;
-     --enable-rbd) printf "%s" -Drbd=enabled ;;
-     --disable-rbd) printf "%s" -Drbd=disabled ;;
-+    --enable-vitastor) printf "%s" -Dvitastor=enabled ;;
-+    --disable-vitastor) printf "%s" -Dvitastor=disabled ;;
-     --enable-rdma) printf "%s" -Drdma=enabled ;;
-     --disable-rdma) printf "%s" -Drdma=disabled ;;
-     --enable-relocatable) printf "%s" -Drelocatable=true ;;
+-static int coroutine_fn vitastor_co_block_status(
+-    BlockDriverState *bs, bool want_zero, int64_t offset, int64_t bytes,
+-    int64_t *pnum, int64_t *map, BlockDriverState **file)
++static int coroutine_fn vitastor_co_block_status(BlockDriverState *bs,
++#if QEMU_VERSION_MAJOR > 10 || QEMU_VERSION_MAJOR == 10 && QEMU_VERSION_MINOR >= 1
++    unsigned int mode,
++#else
++    bool want_zero,
++#endif
++    int64_t offset, int64_t bytes, int64_t *pnum, int64_t *map, BlockDriverState **file)
+ {
+     // Allocated => return BDRV_BLOCK_DATA|BDRV_BLOCK_OFFSET_VALID
+     // Not allocated => return 0
+     // Error => return -errno
+     // Set pnum to length of the extent, `*map` = `offset`, `*file` = `bs`
++#if QEMU_VERSION_MAJOR > 10 || QEMU_VERSION_MAJOR == 10 && QEMU_VERSION_MINOR >= 1
++    int want_zero = (mode == BDRV_WANT_PRECISE);
++#endif
+     VitastorRPC task;
+     VitastorClient *client = bs->opaque;
+     uint64_t inode = client->watch ? vitastor_c_inode_get_num(client->watch) : client->inode;

patches/qemu-10.2-vitastor.patch

@@ -1,172 +0,0 @@
-diff --git a/block/meson.build b/block/meson.build
-index 34b1b2a306..24ca0f1e52 100644
---- a/block/meson.build
-+++ b/block/meson.build
-@@ -114,6 +114,7 @@ foreach m : [
-   [libnfs, 'nfs', files('nfs.c')],
-   [libssh, 'ssh', files('ssh.c')],
-   [rbd, 'rbd', files('rbd.c')],
-+  [vitastor, 'vitastor', files('vitastor.c')],
- ]
-   if m[0].found()
-     module_ss = ss.source_set()
-diff --git a/meson.build b/meson.build
-index d9293294d8..776a5becc6 100644
---- a/meson.build
-+++ b/meson.build
-@@ -1665,6 +1665,26 @@ if not get_option('rbd').auto() or have_block
-   endif
- endif
- 
-+vitastor = not_found
-+if not get_option('vitastor').auto() or have_block
-+  libvitastor_client = cc.find_library('vitastor_client', has_headers: ['vitastor_c.h'],
-+    required: get_option('vitastor'))
-+  if libvitastor_client.found()
-+    if cc.links('''
-+      #include <vitastor_c.h>
-+      int main(void) {
-+        vitastor_c_create_qemu(0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0);
-+        return 0;
-+      }''', dependencies: libvitastor_client)
-+      vitastor = declare_dependency(dependencies: libvitastor_client)
-+    elif get_option('vitastor').enabled()
-+      error('could not link libvitastor_client')
-+    else
-+      warning('could not link libvitastor_client, disabling')
-+    endif
-+  endif
-+endif
-+
- glusterfs = not_found
- glusterfs_ftruncate_has_stat = false
- glusterfs_iocb_has_stat = false
-@@ -2509,6 +2529,7 @@ endif
- config_host_data.set('CONFIG_OPENGL', opengl.found())
- config_host_data.set('CONFIG_PLUGIN', get_option('plugins'))
- config_host_data.set('CONFIG_RBD', rbd.found())
-+config_host_data.set('CONFIG_VITASTOR', vitastor.found())
- config_host_data.set('CONFIG_RDMA', rdma.found())
- config_host_data.set('CONFIG_RELOCATABLE', get_option('relocatable'))
- config_host_data.set('CONFIG_SAFESTACK', get_option('safe_stack'))
-@@ -4948,6 +4969,7 @@ summary_info += {'fdt support':       fdt_opt == 'internal' ? 'internal' : fdt}
- summary_info += {'libcap-ng support': libcap_ng}
- summary_info += {'bpf support':       libbpf}
- summary_info += {'rbd support':       rbd}
-+summary_info += {'vitastor support':  vitastor}
- summary_info += {'smartcard support': cacard}
- summary_info += {'U2F support':       u2f}
- summary_info += {'libusb':            libusb}
-diff --git a/meson_options.txt b/meson_options.txt
-index 2836156257..148086cc6f 100644
---- a/meson_options.txt
-+++ b/meson_options.txt
-@@ -206,6 +206,8 @@ option('pvg', type: 'feature', value: 'auto',
-        description: 'macOS paravirtualized graphics support')
- option('rbd', type : 'feature', value : 'auto',
-        description: 'Ceph block device driver')
-+option('vitastor', type : 'feature', value : 'auto',
-+       description: 'Vitastor block device driver')
- option('opengl', type : 'feature', value : 'auto',
-        description: 'OpenGL support')
- option('rdma', type : 'feature', value : 'auto',
-diff --git a/qapi/block-core.json b/qapi/block-core.json
-index b82af74256..f25a6f5ce8 100644
---- a/qapi/block-core.json
-+++ b/qapi/block-core.json
-@@ -3351,7 +3351,7 @@
-             'parallels', 'preallocate', 'qcow', 'qcow2', 'qed', 'quorum',
-             'raw', 'rbd',
-             { 'name': 'replication', 'if': 'CONFIG_REPLICATION' },
--            'ssh', 'throttle', 'vdi', 'vhdx',
-+            'ssh', 'throttle', 'vdi', 'vhdx', 'vitastor',
-             { 'name': 'virtio-blk-vfio-pci', 'if': 'CONFIG_BLKIO' },
-             { 'name': 'virtio-blk-vhost-user', 'if': 'CONFIG_BLKIO' },
-             { 'name': 'virtio-blk-vhost-vdpa', 'if': 'CONFIG_BLKIO' },
-@@ -4434,6 +4434,28 @@
-             '*key-secret': 'str',
-             '*server': ['InetSocketAddressBase'] } }
- 
-+##
-+# @BlockdevOptionsVitastor:
-+#
-+# Driver specific block device options for vitastor
-+#
-+# @image:       Image name
-+# @inode:       Inode number
-+# @pool:        Pool ID
-+# @size:        Desired image size in bytes
-+# @config-path: Path to Vitastor configuration
-+# @etcd-host:   etcd connection address(es)
-+# @etcd-prefix: etcd key/value prefix
-+##
-+{ 'struct': 'BlockdevOptionsVitastor',
-+  'data': { '*inode': 'uint64',
-+            '*pool': 'uint64',
-+            '*size': 'uint64',
-+            '*image': 'str',
-+            '*config-path': 'str',
-+            '*etcd-host': 'str',
-+            '*etcd-prefix': 'str' } }
-+
- ##
- # @ReplicationMode:
- #
-@@ -4902,6 +4924,7 @@
-       'throttle':   'BlockdevOptionsThrottle',
-       'vdi':        'BlockdevOptionsGenericFormat',
-       'vhdx':       'BlockdevOptionsGenericFormat',
-+      'vitastor':   'BlockdevOptionsVitastor',
-       'virtio-blk-vfio-pci':
-                     { 'type': 'BlockdevOptionsVirtioBlkVfioPci',
-                       'if': 'CONFIG_BLKIO' },
-@@ -5376,6 +5399,20 @@
-             '*cluster-size' :   'size',
-             '*encrypt' :        'RbdEncryptionCreateOptions' } }
- 
-+##
-+# @BlockdevCreateOptionsVitastor:
-+#
-+# Driver specific image creation options for Vitastor.
-+#
-+# @location: Where to store the new image file.  This location cannot
-+#     point to a snapshot.
-+#
-+# @size: Size of the virtual disk in bytes
-+##
-+{ 'struct': 'BlockdevCreateOptionsVitastor',
-+  'data': { 'location':         'BlockdevOptionsVitastor',
-+            'size':             'size' } }
-+
- ##
- # @BlockdevVmdkSubformat:
- #
-@@ -5598,6 +5635,7 @@
-       'ssh':            'BlockdevCreateOptionsSsh',
-       'vdi':            'BlockdevCreateOptionsVdi',
-       'vhdx':           'BlockdevCreateOptionsVhdx',
-+      'vitastor':       'BlockdevCreateOptionsVitastor',
-       'vmdk':           'BlockdevCreateOptionsVmdk',
-       'vpc':            'BlockdevCreateOptionsVpc'
-   } }
-diff --git a/scripts/meson-buildoptions.sh b/scripts/meson-buildoptions.sh
-index 3d0d132344..65ee8c855e 100644
---- a/scripts/meson-buildoptions.sh
-+++ b/scripts/meson-buildoptions.sh
-@@ -177,6 +177,7 @@ meson_options_help() {
-   printf "%s\n" '  qga-vss         build QGA VSS support (broken with MinGW)'
-   printf "%s\n" '  qpl             Query Processing Library support'
-   printf "%s\n" '  rbd             Ceph block device driver'
-+  printf "%s\n" '  vitastor        Vitastor block device driver'
-   printf "%s\n" '  rdma            Enable RDMA-based migration'
-   printf "%s\n" '  replication     replication support'
-   printf "%s\n" '  rust            Rust support'
-@@ -464,6 +465,8 @@ _meson_option_parse() {
-     --disable-qpl) printf "%s" -Dqpl=disabled ;;
-     --enable-rbd) printf "%s" -Drbd=enabled ;;
-     --disable-rbd) printf "%s" -Drbd=disabled ;;
-+    --enable-vitastor) printf "%s" -Dvitastor=enabled ;;
-+    --disable-vitastor) printf "%s" -Dvitastor=disabled ;;
-     --enable-rdma) printf "%s" -Drdma=enabled ;;
-     --disable-rdma) printf "%s" -Drdma=disabled ;;
-     --enable-relocatable) printf "%s" -Drelocatable=true ;;

rpm/build-tarball.sh

@@ -21,7 +21,7 @@ rpmbuild -bp fio.spec
 cd $VITASTOR
 VER=$(grep ^Version: rpm/vitastor-$REL.spec | awk '{print $2}')
 rm -rf fio
-ln -s $(ls -d ~/rpmbuild/BUILD/fio*/ | grep -v SPECPARTS) fio
+ln -s ~/rpmbuild/BUILD/fio*/ fio
 sh copy-fio-includes.sh
 rm fio
 mv fio-copy fio

rpm/vitastor-el10.Dockerfile

@@ -1,17 +0,0 @@
-# Build packages for AlmaLinux 10 inside a container
-# cd ..
-# docker pull --platform=linux/amd64/v2 quay.io/almalinuxorg/almalinux:10
-# docker build -t vitastor-buildenv:el10 -f rpm/vitastor-el10.Dockerfile .
-# docker run -i --rm -v ./:/root/vitastor vitastor-buildenv:el10 /root/vitastor/rpm/vitastor-build.sh
-
-FROM quay.io/almalinuxorg/almalinux:10
-
-WORKDIR /root
-
-RUN sed -i 's/enabled=0/enabled=1/' /etc/yum.repos.d/*.repo
-RUN dnf -y install epel-release dnf-plugins-core
-RUN dnf -y install https://vitastor.io/rpms/centos/10/vitastor-release-1.0-1.el10.noarch.rpm
-RUN dnf -y install gcc-c++ gperftools-devel fio nodejs rpm-build jerasure-devel isa-l-devel gf-complete-devel rdma-core-devel cmake libnl3-devel
-RUN dnf download --source fio
-RUN rpm --nomd5 -i fio*.src.rpm
-RUN cd ~/rpmbuild/SPECS && dnf builddep -y --spec fio.spec

rpm/vitastor-el10.spec

@@ -1,198 +0,0 @@
-Name:           vitastor
-Version:        3.0.12
-Release:        1%{?dist}
-Summary:        Vitastor, a fast software-defined clustered block storage
-
-License:        Vitastor Network Public License 1.1
-URL:            https://vitastor.io/
-Source0:        vitastor-3.0.12.el10.tar.gz
-
-BuildRequires:  gperftools-devel
-BuildRequires:  gcc-c++
-BuildRequires:  nodejs >= 10
-BuildRequires:  jerasure-devel
-BuildRequires:  isa-l-devel
-BuildRequires:  gf-complete-devel
-BuildRequires:  rdma-core-devel
-BuildRequires:  cmake
-BuildRequires:  libnl3-devel
-Requires:       vitastor-osd = %{version}-%{release}
-Requires:       vitastor-mon = %{version}-%{release}
-Requires:       vitastor-client = %{version}-%{release}
-Requires:       vitastor-client-devel = %{version}-%{release}
-Requires:       vitastor-fio = %{version}-%{release}
-
-%description
-Vitastor is a small, simple and fast clustered block storage (storage for VM drives),
-architecturally similar to Ceph which means strong consistency, primary-replication,
-symmetric clustering and automatic data distribution over any number of drives of any
-size with configurable redundancy (replication or erasure codes/XOR).
-
-
-%package -n vitastor-osd
-Summary:        Vitastor - OSD
-Requires:       vitastor-client = %{version}-%{release}
-Requires:       util-linux
-Requires:       parted
-
-
-%description -n vitastor-osd
-Vitastor object storage daemon, i.e. server program that stores data.
-
-
-%package -n vitastor-mon
-Summary:        Vitastor - monitor
-Requires:       nodejs >= 10
-Requires:       lpsolve
-
-
-%description -n vitastor-mon
-Vitastor monitor, i.e. server program responsible for watching cluster state and
-scheduling cluster-level operations.
-
-
-%package -n vitastor-client
-Summary:        Vitastor - client
-
-
-%description -n vitastor-client
-Vitastor client library and command-line interface.
-
-
-%package -n vitastor-client-devel
-Summary:        Vitastor - development files
-Group:          Development/Libraries
-Requires:       vitastor-client = %{version}-%{release}
-
-
-%description -n vitastor-client-devel
-Vitastor library headers for development.
-
-
-%package -n vitastor-fio
-Summary:        Vitastor - fio drivers
-Group:          Development/Libraries
-Requires:       vitastor-client = %{version}-%{release}
-Requires:       fio = 3.36-5.el10
-
-
-%description -n vitastor-fio
-Vitastor fio drivers for benchmarking.
-
-
-%package -n vitastor-opennebula
-Summary:        Vitastor for OpenNebula
-Group:          Development/Libraries
-Requires:       vitastor-client
-Requires:       jq
-Requires:       python3-lxml
-Requires:       patch
-Requires:       qemu-kvm-block-vitastor
-
-
-%description -n vitastor-opennebula
-Vitastor storage plugin for OpenNebula.
-
-
-%prep
-%setup -q
-
-
-%build
-%cmake
-%cmake_build
-
-
-%install
-rm -rf $RPM_BUILD_ROOT
-%cmake_install
-cd mon
-npm install --production
-cd ..
-mkdir -p %buildroot/usr/lib/vitastor
-cp -r mon %buildroot/usr/lib/vitastor
-mv %buildroot/usr/lib/vitastor/mon/scripts/make-etcd %buildroot/usr/lib/vitastor/mon/
-mkdir -p %buildroot/lib/systemd/system
-cp mon/scripts/vitastor.target mon/scripts/vitastor-mon.service mon/scripts/vitastor-osd@.service %buildroot/lib/systemd/system
-mkdir -p %buildroot/lib/udev/rules.d
-cp mon/scripts/90-vitastor.rules %buildroot/lib/udev/rules.d
-mkdir -p %buildroot/var/lib/one
-cp -r opennebula/remotes %buildroot/var/lib/one
-cp opennebula/install.sh %buildroot/var/lib/one/remotes/datastore/vitastor/
-mkdir -p %buildroot/etc/
-cp -r opennebula/sudoers.d %buildroot/etc/
-
-
-%files
-%doc GPL-2.0.txt VNPL-1.1.txt README.md README-ru.md
-
-
-%files -n vitastor-osd
-%_bindir/vitastor-osd
-%_bindir/vitastor-disk
-%_bindir/vitastor-dump-journal
-/lib/systemd/system/vitastor-osd@.service
-/lib/systemd/system/vitastor.target
-/lib/udev/rules.d/90-vitastor.rules
-
-
-%pre -n vitastor-osd
-groupadd -r -f vitastor 2>/dev/null ||:
-useradd -r -g vitastor -s /sbin/nologin -c "Vitastor daemons" -M -d /nonexistent vitastor 2>/dev/null ||:
-install -o vitastor -g vitastor -d /var/log/vitastor
-mkdir -p /etc/vitastor
-
-
-%files -n vitastor-mon
-/usr/lib/vitastor/mon
-/lib/systemd/system/vitastor-mon.service
-
-
-%pre -n vitastor-mon
-groupadd -r -f vitastor 2>/dev/null ||:
-useradd -r -g vitastor -s /sbin/nologin -c "Vitastor daemons" -M -d /nonexistent vitastor 2>/dev/null ||:
-mkdir -p /etc/vitastor
-mkdir -p /var/lib/vitastor
-chown vitastor:vitastor /var/lib/vitastor
-
-
-%files -n vitastor-client
-%_bindir/vitastor-nbd
-%_bindir/vitastor-ublk
-%_bindir/vitastor-nfs
-%_bindir/vitastor-cli
-%_bindir/vitastor-rm
-%_bindir/vitastor-kv
-%_bindir/vitastor-kv-stress
-%_bindir/vita
-%_libdir/libvitastor_client.so*
-%_libdir/libvitastor_kv.so*
-
-
-%files -n vitastor-client-devel
-%_includedir/vitastor_c.h
-%_includedir/vitastor_kv.h
-%_libdir/pkgconfig
-
-
-%files -n vitastor-fio
-%_libdir/libfio_vitastor.so
-%_libdir/libfio_vitastor_blk.so
-%_libdir/libfio_vitastor_sec.so
-
-
-%files -n vitastor-opennebula
-/var/lib/one
-/etc/sudoers.d/opennebula-vitastor
-
-
-%triggerin -n vitastor-opennebula -- opennebula
-[ $2 = 0 ] || exit 0
-/var/lib/one/remotes/datastore/vitastor/install.sh
-
-
-# Turn off the brp-python-bytecompile script
-%global __os_install_post %(echo '%{__os_install_post}' | sed -e 's!/usr/lib[^[:space:]]*/brp-python-bytecompile[[:space:]].*$!!g')
-
-
-%changelog

rpm/vitastor-el7.spec

@@ -1,11 +1,11 @@
 Name:           vitastor
-Version:        3.0.12
+Version:        2.4.3
 Release:        1%{?dist}
 Summary:        Vitastor, a fast software-defined clustered block storage
 
 License:        Vitastor Network Public License 1.1
 URL:            https://vitastor.io/
-Source0:        vitastor-3.0.12.el7.tar.gz
+Source0:        vitastor-2.4.3.el7.tar.gz
 
 BuildRequires:  gperftools-devel
 BuildRequires:  devtoolset-9-gcc-c++
@@ -171,6 +171,7 @@ chown vitastor:vitastor /var/lib/vitastor
 %_bindir/vitastor-kv
 %_bindir/vitastor-kv-stress
 %_bindir/vita
+%_libdir/libvitastor_blk.so*
 %_libdir/libvitastor_client.so*
 %_libdir/libvitastor_kv.so*
 

rpm/vitastor-el8.spec

@@ -1,11 +1,11 @@
 Name:           vitastor
-Version:        3.0.12
+Version:        2.4.3
 Release:        1%{?dist}
 Summary:        Vitastor, a fast software-defined clustered block storage
 
 License:        Vitastor Network Public License 1.1
 URL:            https://vitastor.io/
-Source0:        vitastor-3.0.12.el8.tar.gz
+Source0:        vitastor-2.4.3.el8.tar.gz
 
 BuildRequires:  gperftools-devel
 BuildRequires:  gcc-toolset-9-gcc-c++
@@ -168,6 +168,7 @@ chown vitastor:vitastor /var/lib/vitastor
 %_bindir/vitastor-kv
 %_bindir/vitastor-kv-stress
 %_bindir/vita
+%_libdir/libvitastor_blk.so*
 %_libdir/libvitastor_client.so*
 %_libdir/libvitastor_kv.so*
 

rpm/vitastor-el9.spec

@@ -1,11 +1,11 @@
 Name:           vitastor
-Version:        3.0.12
+Version:        2.4.3
 Release:        1%{?dist}
 Summary:        Vitastor, a fast software-defined clustered block storage
 
 License:        Vitastor Network Public License 1.1
 URL:            https://vitastor.io/
-Source0:        vitastor-3.0.12.el9.tar.gz
+Source0:        vitastor-2.4.3.el9.tar.gz
 
 BuildRequires:  gperftools-devel
 BuildRequires:  gcc-c++
@@ -165,6 +165,7 @@ chown vitastor:vitastor /var/lib/vitastor
 %_bindir/vitastor-kv
 %_bindir/vitastor-kv-stress
 %_bindir/vita
+%_libdir/libvitastor_blk.so*
 %_libdir/libvitastor_client.so*
 %_libdir/libvitastor_kv.so*
 

src/CMakeLists.txt

@@ -1,8 +1,9 @@
-cmake_minimum_required(VERSION 2.8...3.30)
+cmake_minimum_required(VERSION 2.8.12)
 
 project(vitastor)
 
 include(GNUInstallDirs)
+include(CTest)
 include(CheckIncludeFile)
 
 find_package(PkgConfig)
@@ -20,7 +21,7 @@ if("${CMAKE_INSTALL_PREFIX}" MATCHES "^/usr/local/?$")
 endif()
 set(ENABLE_COVERAGE false CACHE BOOL "Enable code coverage")
 
-add_definitions(-DVITASTOR_VERSION="3.0.12")
+add_definitions(-DVITASTOR_VERSION="2.4.3")
 add_definitions(-D_GNU_SOURCE -D_LARGEFILE64_SOURCE -D_FILE_OFFSET_BITS=64 -Wall -Wno-sign-compare -Wno-comment -Wno-parentheses -Wno-pointer-arith -fdiagnostics-color=always -fno-omit-frame-pointer -fvisibility=hidden -I ${CMAKE_SOURCE_DIR}/src)
 add_link_options(-fno-omit-frame-pointer)
 if (${WITH_ASAN})

src/blockstore/CMakeLists.txt

@@ -1,4 +1,4 @@
-cmake_minimum_required(VERSION 2.8...3.30)
+cmake_minimum_required(VERSION 2.8.12)
 
 project(vitastor)
 

src/blockstore/blockstore.h

@@ -183,11 +183,6 @@ public:
     // Update configuration
     virtual void parse_config(blockstore_config_t & config) = 0;
 
-    // Reshard database for a pool in chunks
-    // MUST be called only when nobody makes any modifications to the DB for this pool
-    virtual void* reshard_start(pool_id_t pool, uint32_t pg_count, uint32_t pg_stripe_size, uint64_t chunk_limit) = 0;
-    virtual bool reshard_continue(void *reshard_state, uint64_t chunk_limit) = 0;
-
     // Event loop
     virtual void loop() = 0;
 
@@ -228,9 +223,4 @@ public:
     virtual uint64_t get_journal_size() = 0;
 
     virtual uint32_t get_bitmap_granularity() = 0;
-
-    virtual uint64_t get_live_entries() = 0;
-    virtual uint64_t get_live_memory() = 0;
-    virtual uint64_t get_garbage_entries() = 0;
-    virtual uint64_t get_garbage_memory() = 0;
 };

src/blockstore/blockstore_disk.cpp

@@ -48,12 +48,11 @@ void blockstore_disk_t::parse_config(std::map<std::string, std::string> & config
     disk_alignment = parse_size(config["disk_alignment"]);
     journal_block_size = parse_size(config["journal_block_size"]);
     meta_block_size = parse_size(config["meta_block_size"]);
+    meta_block_target_free_space = parse_size(config["meta_block_target_free_space"]);
     bitmap_granularity = parse_size(config["bitmap_granularity"]);
     meta_format = stoull_full(config["meta_format"]);
     atomic_write_size = (config.find("atomic_write_size") != config.end()
         ? parse_size(config["atomic_write_size"]) : 4096);
-    use_atomic_flag = config.find("use_atomic_flag") != config.end() &&
-        (config["use_atomic_flag"] == "true" || config["use_atomic_flag"] == "1" || config["use_atomic_flag"] == "yes");
     if (config.find("data_io") == config.end() &&
         config.find("meta_io") == config.end() &&
         config.find("journal_io") == config.end())
@@ -94,9 +93,6 @@ void blockstore_disk_t::parse_config(std::map<std::string, std::string> & config
     csum_block_size = parse_size(config["csum_block_size"]);
     discard_on_start = config.find("discard_on_start") != config.end() &&
         (config["discard_on_start"] == "true" || config["discard_on_start"] == "1" || config["discard_on_start"] == "yes");
-    gc_on_start = config.find("gc_on_start") == config.end() ||
-        (config["gc_on_start"] == "true" || config["gc_on_start"] == "1" || config["gc_on_start"] == "yes");
-    skip_double_claim = (config["skip_double_claim"] == "true" || config["skip_double_claim"] == "1" || config["skip_double_claim"] == "yes");
     min_discard_size = parse_size(config["min_discard_size"]);
     if (!min_discard_size)
         min_discard_size = 1024*1024;
@@ -158,6 +154,14 @@ void blockstore_disk_t::parse_config(std::map<std::string, std::string> & config
     {
         throw std::runtime_error("meta_block_size must not exceed "+std::to_string(MAX_DATA_BLOCK_SIZE));
     }
+    if (!meta_block_target_free_space)
+    {
+        meta_block_target_free_space = 800;
+    }
+    if (meta_block_target_free_space >= meta_block_size)
+    {
+        throw std::runtime_error("meta_block_target_free_space must not exceed "+std::to_string(meta_block_size));
+    }
     if (data_offset % disk_alignment)
     {
         throw std::runtime_error("data_offset must be a multiple of disk_alignment = "+std::to_string(disk_alignment));
@@ -174,10 +178,6 @@ void blockstore_disk_t::parse_config(std::map<std::string, std::string> & config
     {
         throw std::runtime_error("Data block size must be a multiple of sparse write tracking granularity");
     }
-    if (data_block_size / bitmap_granularity < 8)
-    {
-        throw std::runtime_error("Data block size must be at least bitmap_granularity*8");
-    }
     if (!data_csum_type)
     {
         csum_block_size = 0;
@@ -210,6 +210,10 @@ void blockstore_disk_t::parse_config(std::map<std::string, std::string> & config
     {
         throw std::runtime_error("journal_offset must be a multiple of journal_block_size = "+std::to_string(journal_block_size));
     }
+    if (!meta_format)
+    {
+        meta_format = BLOCKSTORE_META_FORMAT_HEAP;
+    }
     if (meta_device == data_device)
     {
         disable_meta_fsync = disable_data_fsync;
@@ -220,7 +224,7 @@ void blockstore_disk_t::parse_config(std::map<std::string, std::string> & config
     }
 }
 
-void blockstore_disk_t::calc_lengths(bool skip_meta_check)
+void blockstore_disk_t::calc_lengths()
 {
     // data
     data_len = data_device_size - data_offset;
@@ -266,13 +270,13 @@ void blockstore_disk_t::calc_lengths(bool skip_meta_check)
     }
     // required metadata size
     block_count = data_len / data_block_size;
-    clean_entry_bitmap_size = (data_block_size / bitmap_granularity + 7) / 8;
+    clean_entry_bitmap_size = data_block_size / bitmap_granularity / 8;
     clean_dyn_size = clean_entry_bitmap_size*2 + (csum_block_size
         ? data_block_size/csum_block_size*(data_csum_type & 0xFF) : 0);
-recalc:
     if (meta_format == BLOCKSTORE_META_FORMAT_HEAP)
     {
-        uint32_t entries_per_block = meta_block_size / (sizeof(heap_big_write_t) + clean_dyn_size);
+        uint32_t entries_per_block = ((meta_block_size-meta_block_target_free_space) /
+            (sizeof(heap_object_t) + sizeof(heap_write_t) + clean_dyn_size));
         min_meta_len = (block_count+entries_per_block-1) / entries_per_block * meta_block_size;
     }
     else if (meta_format == BLOCKSTORE_META_FORMAT_V1)
@@ -280,32 +284,11 @@ recalc:
         clean_entry_size = 24 /*sizeof(clean_disk_entry)*/ + 2*clean_entry_bitmap_size;
         min_meta_len = (1 + (block_count - 1 + meta_block_size / clean_entry_size)
             / (meta_block_size / clean_entry_size)) * meta_block_size;
-        if (!skip_meta_check && meta_area_size < min_meta_len)
-        {
-too_small:
-            throw std::runtime_error("Metadata area is too small, need at least "+std::to_string(min_meta_len)+
-                " bytes, have only "+std::to_string(meta_area_size)+" bytes");
-        }
     }
-    else if (meta_format == BLOCKSTORE_META_FORMAT_V2 || !meta_format)
+    else if (meta_format == BLOCKSTORE_META_FORMAT_V2)
     {
-        meta_format = BLOCKSTORE_META_FORMAT_V2;
         clean_entry_size = 24 /*sizeof(clean_disk_entry)*/ + clean_dyn_size + 4 /*entry_csum*/;
         min_meta_len = (1 + (block_count - 1 + meta_block_size / clean_entry_size) / (meta_block_size / clean_entry_size)) * meta_block_size;
-        if (!skip_meta_check && meta_area_size < min_meta_len)
-        {
-            if (!data_csum_type)
-            {
-                printf("Warning: Using old metadata format without checksums because the new format"
-                    " doesn't fit into provided area (%ju bytes required, %ju bytes available)\n", min_meta_len, meta_area_size);
-                meta_format = BLOCKSTORE_META_FORMAT_V1;
-                goto recalc;
-            }
-            else
-            {
-                goto too_small;
-            }
-        }
     }
     else
     {

src/blockstore/blockstore_disk.h

@@ -26,18 +26,18 @@ class allocator_t;
 struct blockstore_disk_t
 {
     std::string data_device, meta_device, journal_device;
-    uint64_t data_block_size;
+    uint32_t data_block_size;
     uint64_t cfg_journal_size, cfg_data_size;
     // Required write alignment and journal/metadata/data areas' location alignment
     uint32_t disk_alignment = 4096;
     // Journal block size - minimum_io_size of the journal device is the best choice
-    uint64_t journal_block_size = 4096;
+    uint32_t journal_block_size = 4096;
     // Metadata block size - minimum_io_size of the metadata device is the best choice
-    uint64_t meta_block_size = 4096;
+    uint32_t meta_block_size = 4096;
     // Atomic write size of the data block device
     uint32_t atomic_write_size = 4096;
-    // Whether we should set RWF_ATOMIC on atomic writes
-    bool use_atomic_flag = false;
+    // Target free space in metadata blocks
+    uint32_t meta_block_target_free_space = 800;
     // Sparse write tracking granularity. 4 KB is a good choice. Must be a multiple of disk_alignment
     uint32_t bitmap_granularity = 4096;
     // Data checksum type, BLOCKSTORE_CSUM_NONE or BLOCKSTORE_CSUM_CRC32C
@@ -57,18 +57,13 @@ struct blockstore_disk_t
     bool inmemory_journal = true;
     // Data discard granularity and minimum size (for the sake of performance)
     bool discard_on_start = false;
-    // GC on start (new store)
-    bool gc_on_start = true;
-    // Skip double claim conflicts on start (new store, temporary until the bug is found)
-    bool skip_double_claim = false;
     uint64_t min_discard_size = 1024*1024;
     uint64_t discard_granularity = 0;
 
     int meta_fd = -1, data_fd = -1, journal_fd = -1;
-    uint64_t meta_offset = 0, meta_device_sect = 0, meta_device_size = 0, meta_area_size = 0, min_meta_len = 0;
-    uint64_t data_offset = 0, data_device_sect = 0, data_device_size = 0, data_len = 0;
-    uint64_t journal_offset = 0, journal_device_sect = 0, journal_device_size = 0, journal_len = 0;
-    uint64_t meta_format = 0;
+    uint64_t meta_offset, meta_device_sect, meta_device_size, meta_area_size, min_meta_len, meta_format = 0;
+    uint64_t data_offset, data_device_sect, data_device_size, data_len;
+    uint64_t journal_offset, journal_device_sect, journal_device_size, journal_len;
 
     uint64_t block_count = 0;
     uint32_t clean_entry_bitmap_size = 0;
@@ -80,7 +75,7 @@ struct blockstore_disk_t
     void open_data();
     void open_meta();
     void open_journal();
-    void calc_lengths(bool skip_meta_check = false);
+    void calc_lengths();
     void check_lengths();
     void close_all();
     int trim_data(std::function<bool(uint64_t)> is_free);

src/blockstore/blockstore_flush.cpp

@@ -18,18 +18,12 @@ journal_flusher_t::journal_flusher_t(blockstore_impl_t *bs)
     this->cur_flusher_count = bs->min_flusher_count;
     this->target_flusher_count = bs->min_flusher_count;
     active_flushers = 0;
+    advance_lsn_counter = 0;
     co = new journal_flusher_co[max_flusher_count];
     for (int i = 0; i < max_flusher_count; i++)
     {
         co[i].co_id = i;
         co[i].bs = bs;
-        co[i].new_bmp = (uint8_t*)malloc_or_die(3*bs->dsk.clean_entry_bitmap_size);
-        co[i].new_ext_bmp = co[i].new_bmp + bs->dsk.clean_entry_bitmap_size;
-        co[i].punch_bmp = co[i].new_bmp + 2*bs->dsk.clean_entry_bitmap_size;
-        if (bs->dsk.csum_block_size > 0)
-        {
-            co[i].new_csums = (uint8_t*)malloc_or_die(bs->dsk.data_block_size / bs->dsk.csum_block_size * (bs->dsk.data_csum_type & 0xFF));
-        }
         co[i].flusher = this;
     }
 }
@@ -60,18 +54,6 @@ journal_flusher_t::~journal_flusher_t()
 
 journal_flusher_co::~journal_flusher_co()
 {
-    if (new_csums)
-    {
-        free(new_csums);
-        new_csums = NULL;
-    }
-    if (new_bmp)
-    {
-        free(new_bmp);
-        new_bmp = NULL;
-    }
-    new_ext_bmp = NULL;
-    punch_bmp = NULL;
     free_buffers();
 }
 
@@ -80,6 +62,11 @@ int journal_flusher_t::get_syncing_buffer()
     return syncing_buffer;
 }
 
+uint64_t journal_flusher_t::get_compact_counter()
+{
+    return compact_counter;
+}
+
 bool journal_flusher_t::is_active()
 {
     return active_flushers > 0;
@@ -99,10 +86,9 @@ void journal_flusher_t::release_trim()
 void journal_flusher_t::dump_diagnostics()
 {
     printf(
-        "Compaction queue: %u items + %u future, data: %ju/%ju blocks used, buffer: %ju/%ju bytes used, meta: %ju/%ju bytes used, %u/%ju blocks nearfull\n",
-        bs->heap->get_compact_queue_size(), bs->heap->get_to_compact_count(),
+        "Compaction queue: %u items, data: %ju/%ju blocks used, meta: %ju/%ju bytes used, %u/%ju blocks nearfull\n",
+        bs->heap->get_to_compact_count(),
         bs->heap->get_data_used_space()/bs->dsk.data_block_size, bs->dsk.block_count,
-        bs->heap->get_buffer_area_used_space(), bs->dsk.journal_len,
         bs->heap->get_meta_used_space(), bs->heap->get_meta_total_space(),
         bs->heap->get_meta_nearfull_blocks(), bs->dsk.meta_area_size/bs->dsk.meta_block_size-1
     );
@@ -127,12 +113,8 @@ void journal_flusher_t::loop()
         }
     }
     int prev_active = active_flushers;
-    for (int i = 0; (active_flushers > 0 || force_start > 0 ||
-        bs->heap->get_compact_queue_size() > bs->flusher_start_threshold ||
-        i == 0 && bs->intent_write_counter >= bs->journal_trim_interval) && i < cur_flusher_count; i++)
-    {
+    for (int i = 0; (active_flushers > 0 || force_start > 0 || bs->heap->get_to_compact_count() > bs->flusher_start_threshold) && i < cur_flusher_count; i++)
         co[i].loop();
-    }
     if (prev_active && !active_flushers && force_start > 0)
         bs->ringloop->wakeup();
 }
@@ -177,35 +159,19 @@ bool journal_flusher_co::loop()
     else if (wait_state == 22) goto resume_22;
     else if (wait_state == 23) goto resume_23;
     else if (wait_state == 24) goto resume_24;
-    else if (wait_state == 25) goto resume_25;
 resume_0:
     wait_state = 0;
     wait_count = 0;
     cur_oid = {};
     res = bs->heap->get_next_compact(cur_oid);
-    if ((bs->intent_write_counter >= bs->journal_trim_interval) && co_id == 0)
-    {
-        // Advance fsynced_lsn every <journal_trim_interval> intent writes
-        bs->intent_write_counter = 0;
-resume_17:
-resume_18:
-resume_19:
-        if (!fsync_buffer(17))
-            return false;
-resume_20:
-resume_21:
-        if (!trim_lsn(20))
-            return false;
-    }
     if (res == ENOENT && flusher->force_start > 0 && co_id == 0 &&
-        (!bs->dsk.disable_journal_fsync || !bs->dsk.disable_meta_fsync || !bs->dsk.disable_data_fsync))
+        (!bs->dsk.disable_journal_fsync || !bs->dsk.disable_meta_fsync))
     {
-        // When under pressure, do an additional fsync to force entries to be marked compactable
         flusher->active_flushers++;
-resume_14:
-resume_15:
-resume_16:
-        if (!fsync_buffer(14))
+resume_21:
+resume_22:
+        res = fsync_buffer(21);
+        if (!res)
         {
             return false;
         }
@@ -214,105 +180,62 @@ resume_16:
     }
     if (res == ENOENT)
     {
+        if (co_id == 0 && flusher->force_start > 0)
+        {
+            flusher->active_flushers++;
+resume_16:
+resume_17:
+resume_18:
+resume_19:
+resume_20:
+            if (!trim_lsn(16))
+                return false;
+            flusher->active_flushers--;
+        }
         cur_oid = {};
         wait_state = 0;
         return true;
     }
-    if (flusher->flushing.find(cur_oid) != flusher->flushing.end())
+    for (int i = 0; i < flusher->cur_flusher_count; i++)
     {
-        for (int i = 0; i < flusher->cur_flusher_count; i++)
+        if (i != co_id && flusher->co[i].cur_oid == cur_oid)
         {
-            if (i != co_id && flusher->co[i].cur_oid == cur_oid)
-            {
-                // Already flushing it
-                flusher->co[i].should_repeat = true;
-                goto resume_0;
-            }
+            // Already flushing it
+            flusher->co[i].should_repeat = true;
+            goto resume_0;
         }
-        assert(false);
     }
-    flusher->flushing.insert(cur_oid);
 resume_1:
-    wait_state = 1;
     should_repeat = false;
-    cur_obj = bs->heap->lock_and_read_entry(cur_oid);
+    cur_obj = bs->heap->lock_and_read_entry(cur_oid, copy_id);
     if (!cur_obj)
     {
         // Object does not exist
-        flusher->flushing.erase(cur_oid);
         goto resume_0;
     }
+    cur_version = cur_obj->get_writes()->version;
+    // Find the range to compact
+    compact_lsn = bs->heap->get_fsynced_lsn();
+    bs->heap->get_compact_range(cur_obj, compact_lsn, &begin_wr, &end_wr);
+    if (!begin_wr)
+    {
+        // Nothing to flush
+        bs->heap->unlock_entry(cur_oid, copy_id);
+        goto resume_0;
+    }
+    assert(!end_wr->next() && end_wr->entry_type == (BS_HEAP_BIG_WRITE|BS_HEAP_STABLE));
+    clean_loc = end_wr->big_location(bs->heap);
+    if (bs->log_level > 10)
+        printf("Compacting %jx:%jx l%ju .. l%ju (last l%ju)\n", cur_oid.inode, cur_oid.stripe, end_wr->lsn, begin_wr->lsn, compact_lsn);
+    flusher->active_flushers++;
     // Scan versions to flush
     free_buffers();
     copy_count = 0;
-    fsynced_lsn = bs->heap->get_fsynced_lsn();
-    bitmap_copied = false;
-    memset(new_bmp, 0, bs->dsk.clean_entry_bitmap_size);
-    csum_copy.clear();
-    compact_info = bs->heap->iterate_compaction(cur_obj, fsynced_lsn, flusher->force_start, [&](heap_entry_t *wr)
+    for (auto wr = begin_wr; wr != end_wr; wr = wr->next())
     {
-        if (!bitmap_copied)
-        {
-            memcpy(new_ext_bmp, wr->get_ext_bitmap(bs->heap), bs->dsk.clean_entry_bitmap_size);
-            bitmap_copied = true;
-        }
-        bitmap_set(new_bmp, wr->small().offset, wr->small().len, bs->dsk.bitmap_granularity);
-        if (bs->dsk.csum_block_size && bs->dsk.csum_block_size <= bs->dsk.bitmap_granularity)
-        {
-            csum_copy.push_back(wr);
-        }
-        if (wr->type() == BS_HEAP_SMALL_WRITE ||
-            wr->type() == BS_HEAP_INTENT_WRITE && bs->dsk.csum_block_size > bs->dsk.bitmap_granularity)
-        {
-            bs->prepare_read(read_vec, cur_obj, wr, 0, bs->dsk.data_block_size,
-                wr->type() == BS_HEAP_INTENT_WRITE && bs->dsk.csum_block_size > bs->dsk.bitmap_granularity && !bs->perfect_csum_update
-                ? COPY_BUF_SKIP_CSUM : 0);
-        }
-    });
-    if (!compact_info.compact_lsn)
-    {
-        // Flushing is aborted
-        flusher->flushing.erase(cur_oid);
-        bs->heap->unlock_entry(cur_oid);
-        goto resume_0;
+        bs->prepare_read(read_vec, cur_obj, wr, 0, bs->dsk.data_block_size);
+        copy_count++;
     }
-    flusher->active_flushers++;
-    for (i = 0; i < read_vec.size(); i++)
-    {
-        if ((read_vec[i].copy_flags & COPY_BUF_JOURNAL) &&
-            !(read_vec[i].copy_flags & COPY_BUF_COALESCED))
-        {
-            copy_count++;
-        }
-    }
-    if (copy_count > 0 && !bs->dsk.disable_data_fsync)
-    {
-        init_fsync_data();
-    }
-    if (bs->log_level > 10)
-    {
-        printf("Compacting %jx:%jx v%ju..v%ju / l%ju..l%ju (%d writes)\n", cur_oid.inode, cur_oid.stripe,
-            compact_info.clean_wr->version, compact_info.compact_version,
-            compact_info.clean_wr->lsn, compact_info.compact_lsn, copy_count);
-    }
-    mem_or(new_bmp, compact_info.clean_wr->get_int_bitmap(bs->heap), bs->dsk.clean_entry_bitmap_size);
-    if (!bitmap_copied)
-    {
-        memcpy(new_ext_bmp, compact_info.clean_wr->get_ext_bitmap(bs->heap), bs->dsk.clean_entry_bitmap_size);
-        bitmap_copied = true;
-    }
-    if (bs->dsk.csum_block_size && bs->dsk.csum_block_size <= bs->dsk.bitmap_granularity)
-    {
-        memcpy(new_csums, compact_info.clean_wr->get_checksums(bs->heap), bs->dsk.data_block_size/bs->dsk.csum_block_size * (bs->dsk.data_csum_type & 0xFF));
-        for (size_t i = csum_copy.size(); i > 0; i--)
-        {
-            auto wr = csum_copy[i-1];
-            memcpy(new_csums + wr->small().offset/bs->dsk.csum_block_size*(bs->dsk.data_csum_type & 0xFF),
-                wr->get_checksums(bs->heap), wr->small().len/bs->dsk.csum_block_size*(bs->dsk.data_csum_type & 0xFF));
-        }
-        csum_copy.clear();
-    }
-    clean_loc = compact_info.clean_wr->big_location(bs->heap);
     overwrite_start = overwrite_end = 0;
     if (read_vec.size() > 0)
     {
@@ -331,6 +254,7 @@ resume_1:
     }
     // Read buffered data
     cur_obj = NULL;
+    begin_wr = end_wr = NULL;
 resume_2:
 resume_3:
     if (!read_buffered(2))
@@ -348,135 +272,107 @@ resume_3:
         flusher->wanting_meta_fsync--;
     }
     res = check_and_punch_checksums();
-    if (res == ENOENT || res == EDOM)
-    {
-        // Abort compaction
-abort_compact:
-        if (copy_count > 0 && !bs->dsk.disable_data_fsync)
-        {
-            cur_sync->member_count--;
-            if (cur_sync->member_count > 0)
-                bs->ringloop->wakeup();
-        }
-        flusher->flushing.erase(cur_oid);
-        bs->heap->unlock_entry(cur_oid);
-        flusher->active_flushers--;
-        goto resume_0;
-    }
     if (res == EBUSY)
     {
 resume_4:
-        modified_block = UINT32_MAX;
-        res = bs->heap->punch_holes(compact_info.clean_wr, punch_bmp, new_csums, &modified_block);
-        if (res == ENOENT)
-        {
-            // Abort compaction
-            goto abort_compact;
-        }
-        if (res == EAGAIN)
-        {
-            // Retry, block is busy
-            wait_state = 4;
-            return false;
-        }
-        assert(res == 0);
 resume_5:
-resume_6:
-        if (!write_meta_block(5))
+        if (!write_meta_block(4))
         {
             return false;
         }
+resume_6:
 resume_7:
 resume_8:
-resume_9:
-        if (!fsync_meta(7))
+        if (!fsync_meta(6))
         {
             return false;
         }
         res = 0;
     }
+    else if (res == ENOENT || res == EDOM)
+    {
+        // Abort compaction
+        flusher->active_flushers--;
+        goto resume_0;
+    }
     assert(res == 0);
     // Submit data writes
     for (i = 0; i < read_vec.size(); i++)
     {
         if ((read_vec[i].copy_flags & COPY_BUF_JOURNAL) &&
-            !(read_vec[i].copy_flags & COPY_BUF_COALESCED))
+            !(read_vec[i].copy_flags & COPY_BUF_COALESCED) ||
+            (read_vec[i].copy_flags & COPY_BUF_PADDED)) // FIXME Shit, simplify these flags
         {
             assert(read_vec[i].buf);
-            await_sqe(10);
+            await_sqe(9);
             data->iov = (struct iovec){ read_vec[i].buf + (read_vec[i].copy_flags & COPY_BUF_PADDED
                 ? read_vec[i].offset - read_vec[i].disk_offset : 0), (size_t)read_vec[i].len };
             data->callback = simple_callback_w;
-            assert(clean_loc + read_vec[i].offset + data->iov.iov_len <= bs->dsk.block_count*bs->dsk.data_block_size);
             io_uring_prep_writev(sqe, bs->dsk.data_fd, &data->iov, 1, bs->dsk.data_offset + clean_loc + read_vec[i].offset);
             wait_count++;
         }
     }
-resume_11:
+resume_10:
     if (wait_count > 0)
     {
-        wait_state = 11;
+        wait_state = 10;
         return false;
     }
-    if (copy_count > 0 && !bs->dsk.disable_data_fsync)
-    {
-resume_22:
-resume_23:
-resume_24:
-resume_25:
-        if (!fsync_data(22))
-        {
-            return false;
-        }
-    }
     // Lock is only needed to prevent freeing the big_write because we overwrite it...
-    bs->heap->unlock_entry(cur_oid);
+    bs->heap->unlock_entry(cur_oid, copy_id);
     // Mark the object compacted, but don't free and remove small_writes
     // We'll free and remove them only when trimming
     // The only thing we modify here are big_write block checksums if >4k block is used
-    cur_obj = bs->heap->read_entry(cur_oid);
+    cur_obj = bs->heap->read_entry(cur_oid, &modified_block);
     if (!cur_obj)
     {
         // Abort compaction
-        flusher->active_flushers--;
-        flusher->flushing.erase(cur_oid);
         goto resume_0;
     }
     if (!calc_block_checksums())
     {
         // Abort compaction
-        flusher->active_flushers--;
-        flusher->flushing.erase(cur_oid);
         goto resume_0;
     }
-    res = bs->heap->add_compact(cur_obj, compact_info.compact_version, compact_info.compact_lsn, clean_loc,
-        compact_info.do_delete, &modified_block, new_bmp, new_ext_bmp, new_csums);
-    if (res == EBUSY)
+    if (read_to_fill_incomplete)
     {
-        // Abort compaction, object is already overwritten by something else
-        flusher->active_flushers--;
-        flusher->flushing.erase(cur_oid);
-        goto resume_0;
-    }
-    assert(res == 0);
-resume_12:
-resume_13:
-    if (!write_meta_block(12))
-    {
-        return false;
+resume_23:
+resume_24:
+        if (!write_meta_block(23))
+        {
+            return false;
+        }
+        // Recheck the object because it could be invalidated again
+        cur_obj = bs->heap->read_entry(cur_oid, &modified_block);
+        if (!cur_obj)
+        {
+            // Abort compaction
+            goto resume_0;
+        }
     }
+    bs->heap->mark_object_compacted(cur_obj, compact_lsn);
     // Done
     if (bs->log_level > 10)
-    {
-        printf("Compacted %jx:%jx l%ju (%d writes)\n", cur_oid.inode, cur_oid.stripe, compact_info.compact_lsn, copy_count);
-    }
+        printf("Compacted %jx:%jx l%ju (%d writes)\n", cur_oid.inode, cur_oid.stripe, compact_lsn, copy_count);
+    flusher->compact_counter++;
     flusher->active_flushers--;
+    // Advance compacted_lsn every <journal_trim_interval> objects
+    if (co_id == 0 && !((++flusher->advance_lsn_counter) % bs->journal_trim_interval))
+    {
+        flusher->advance_lsn_counter = 0;
+resume_11:
+resume_12:
+resume_13:
+resume_14:
+resume_15:
+        if (!trim_lsn(11))
+            return false;
+    }
     if (should_repeat)
     {
         // Flush the same object again
         goto resume_1;
     }
-    flusher->flushing.erase(cur_oid);
     // All done
     goto resume_0;
 }
@@ -513,11 +409,11 @@ void journal_flusher_co::fill_partial_checksum_blocks()
                 .copy_flags = COPY_BUF_DATA | copy_flags,
                 .offset = blk_begin,
                 .len = blk_end - blk_begin,
-                .disk_loc = clean_loc,
+                .disk_loc = end_wr->big_location(bs->heap),
                 .disk_offset = blk_begin,
                 .disk_len = blk_end - blk_begin,
                 .buf = (uint8_t*)memalign_or_die(MEM_ALIGNMENT, blk_end - blk_begin),
-                .wr = compact_info.clean_wr,
+                .wr_lsn = end_wr->lsn,
             });
         }
         auto & vec = read_vec[read_vec.size()-1];
@@ -555,28 +451,26 @@ int journal_flusher_co::check_and_punch_checksums()
         return 0;
     }
     // Verify data checksums
-    cur_obj = bs->heap->read_entry(cur_oid);
-    if (!cur_obj)
-    {
-        // Object is deleted, abort compaction
-        return ENOENT;
-    }
+    cur_obj = bs->heap->read_locked_entry(cur_oid, copy_id);
     bool csum_ok = true;
     for (int i = 0; i < read_vec.size(); i++)
     {
         auto & vec = read_vec[i];
         if (!(vec.copy_flags & (COPY_BUF_COALESCED|COPY_BUF_ZERO|COPY_BUF_SKIP_CSUM)))
         {
-            uint32_t *csums = (uint32_t*)(vec.wr->get_checksums(bs->heap)
+            heap_write_t *wr = cur_obj->get_writes();
+            while (wr && wr->lsn != vec.wr_lsn)
+                wr = wr->next();
+            assert(wr);
+            uint32_t *csums = (uint32_t*)(wr->get_checksums(bs->heap)
                 + (vec.disk_offset/bs->dsk.csum_block_size)*(bs->dsk.data_csum_type & 0xFF)
-                - ((vec.wr->type() == BS_HEAP_BIG_WRITE || vec.wr->type() == BS_HEAP_BIG_INTENT)
-                    ? 0 : (vec.wr->small().offset/bs->dsk.csum_block_size)*(bs->dsk.data_csum_type & 0xFF)));
+                - ((wr->type() == BS_HEAP_BIG_WRITE) ? 0 : (wr->small().offset/bs->dsk.csum_block_size)*(bs->dsk.data_csum_type & 0xFF)));
             bs->heap->calc_block_checksums(
-                csums, vec.buf, vec.wr->get_int_bitmap(bs->heap), vec.disk_offset, vec.disk_offset+vec.disk_len, false,
+                csums, vec.buf, wr->get_int_bitmap(bs->heap), vec.disk_offset, vec.disk_offset+vec.disk_len, false,
                 [&](uint32_t mismatch_pos, uint32_t expected_csum, uint32_t real_csum)
                 {
                     printf("Checksum mismatch during compaction in object %jx:%jx v%ju, offset 0x%x in %s area at offset 0x%jx: got %08x, expected %08x\n",
-                        cur_oid.inode, cur_oid.stripe, vec.wr->version, mismatch_pos,
+                        cur_oid.inode, cur_oid.stripe, wr->version, mismatch_pos,
                         (vec.copy_flags & COPY_BUF_JOURNAL ? "buffer" : "data"),
                         vec.disk_loc+vec.disk_offset, real_csum, expected_csum);
                     csum_ok = false;
@@ -595,8 +489,20 @@ int journal_flusher_co::check_and_punch_checksums()
         // Nothing to do
         return 0;
     }
-    memcpy(punch_bmp, compact_info.clean_wr->get_int_bitmap(bs->heap), bs->dsk.clean_entry_bitmap_size);
-    memcpy(new_csums, compact_info.clean_wr->get_checksums(bs->heap), bs->dsk.data_block_size/bs->dsk.csum_block_size * (bs->dsk.data_csum_type & 0xFF));
+    cur_obj = bs->heap->read_entry(cur_oid, &modified_block);
+    if (!cur_obj)
+    {
+        // Object is deleted, abort compaction
+        return ENOENT;
+    }
+    bs->heap->get_compact_range(cur_obj, compact_lsn, &begin_wr, &end_wr);
+    if (!begin_wr)
+    {
+        // Object is overwritten, abort compaction
+        return ENOENT;
+    }
+    uint8_t *bmp = end_wr->get_int_bitmap(bs->heap);
+    uint8_t *csums = end_wr->get_checksums(bs->heap);
     // Clear bits
     for (auto & vec: read_vec)
     {
@@ -607,7 +513,7 @@ int journal_flusher_co::check_and_punch_checksums()
         if (!(vec.copy_flags & COPY_BUF_COALESCED) &&
             ((vec.offset % bs->dsk.csum_block_size) || (vec.len % bs->dsk.csum_block_size)))
         {
-            bitmap_clear(punch_bmp, vec.offset, vec.len, bs->dsk.bitmap_granularity);
+            bitmap_clear(bmp, vec.offset, vec.len, bs->dsk.bitmap_granularity);
         }
     }
     // Update partial block checksums
@@ -616,20 +522,42 @@ int journal_flusher_co::check_and_punch_checksums()
         if (vec.copy_flags & COPY_BUF_CSUM_FILL)
         {
             uint32_t csum_off = vec.offset/bs->dsk.csum_block_size * (bs->dsk.data_csum_type & 0xFF);
-            bs->heap->calc_block_checksums((uint32_t*)(new_csums+csum_off), vec.buf, punch_bmp, vec.offset, vec.offset+vec.len, true, NULL);
+            bs->heap->calc_block_checksums((uint32_t*)(csums+csum_off), vec.buf, bmp, vec.offset, vec.offset+vec.len, true, NULL);
         }
     }
-    // Modified, we should punch_holes and then write the block to disk
+    cur_obj->crc32c = cur_obj->calc_crc32c();
+    if (res == ENOENT)
+    {
+        // Object is deleted, abort compaction
+        return ENOENT;
+    }
+    // Modified, we should write the block to disk
+    assert(!res);
     return EBUSY;
 }
 
 bool journal_flusher_co::calc_block_checksums()
 {
-    if (bs->dsk.csum_block_size <= bs->dsk.bitmap_granularity)
+    if (bs->dsk.csum_block_size <= bs->dsk.bitmap_granularity || !read_vec.size())
     {
         return true;
     }
-    memcpy(new_csums, compact_info.clean_wr->get_checksums(bs->heap), bs->dsk.data_block_size/bs->dsk.csum_block_size * (bs->dsk.data_csum_type & 0xFF));
+    bs->heap->get_compact_range(cur_obj, compact_lsn, &begin_wr, &end_wr);
+    if (!begin_wr)
+    {
+        // Object is overwritten, abort compaction
+        return false;
+    }
+    uint8_t *bmp = end_wr->get_int_bitmap(bs->heap);
+    uint8_t *csums = end_wr->get_checksums(bs->heap);
+    // Set bits
+    for (auto & vec: read_vec)
+    {
+        if (!(vec.copy_flags & (COPY_BUF_COALESCED|COPY_BUF_CSUM_FILL)))
+        {
+            bitmap_set(bmp, vec.offset, vec.len, bs->dsk.bitmap_granularity);
+        }
+    }
     // Update block checksums
     size_t i = 0;
     while (i < read_vec.size() && !(read_vec[i].copy_flags & COPY_BUF_CSUM_FILL))
@@ -648,7 +576,7 @@ bool journal_flusher_co::calc_block_checksums()
         assert(!(end % bs->dsk.csum_block_size));
         uint32_t csum_off = start/bs->dsk.csum_block_size * (bs->dsk.data_csum_type & 0xFF);
         bs->heap->calc_block_checksums(
-            (uint32_t*)(new_csums+csum_off), new_bmp, start, end,
+            (uint32_t*)(csums+csum_off), bmp, start, end,
             [&](uint32_t start, uint32_t & len)
             {
                 // O(n^2) search, may be fixed later :-p
@@ -665,6 +593,7 @@ bool journal_flusher_co::calc_block_checksums()
             }, true, NULL
         );
     }
+    cur_obj->crc32c = cur_obj->calc_crc32c();
     return true;
 }
 
@@ -674,15 +603,13 @@ bool journal_flusher_co::write_meta_block(int wait_base)
         goto resume_0;
     else if (wait_state == wait_base+1)
         goto resume_1;
-resume_0:
-    if (bs->ringloop->space_left() < 1)
-    {
-        wait_state = wait_base+0;
-        return 0;
-    }
-    bs->prepare_meta_block_write(modified_block);
+    await_sqe(0);
+    data->iov = (struct iovec){ bs->heap->get_meta_block(modified_block), (size_t)bs->dsk.meta_block_size };
+    data->callback = simple_callback_w;
+    io_uring_prep_writev(sqe, bs->dsk.meta_fd, &data->iov, 1, bs->dsk.meta_offset + (modified_block+1)*bs->dsk.meta_block_size);
+    wait_count++;
 resume_1:
-    if (bs->meta_block_is_pending(modified_block))
+    if (wait_count > 0)
     {
         wait_state = wait_base+1;
         return false;
@@ -732,77 +659,16 @@ resume_1:
     return true;
 }
 
-void journal_flusher_co::init_fsync_data()
-{
-    cur_sync = flusher->data_syncs.begin();
-    if (cur_sync == flusher->data_syncs.end() || cur_sync->ready_count > 0)
-    {
-        cur_sync = flusher->data_syncs.emplace(cur_sync);
-    }
-    cur_sync->member_count++;
-}
-
-bool journal_flusher_co::fsync_data(int wait_base)
-{
-    if (wait_state == wait_base)
-        goto resume_0;
-    else if (wait_state == wait_base+1)
-        goto resume_1;
-    else if (wait_state == wait_base+2)
-        goto resume_2;
-    else if (wait_state == wait_base+3)
-        goto resume_3;
-    cur_sync->ready_count++;
-resume_0:
-    if (cur_sync->ready_count < cur_sync->member_count)
-    {
-        wait_state = wait_base;
-        return false;
-    }
-    if (!cur_sync->sent)
-    {
-        // Sync batch is ready. Do it.
-        await_sqe(1);
-        data->iov = { 0 };
-        data->callback = simple_callback_w;
-        io_uring_prep_fsync(sqe, bs->dsk.data_fd, IORING_FSYNC_DATASYNC);
-        cur_sync->sent = true;
-        wait_count++;
-resume_2:
-        if (wait_count > 0)
-        {
-            wait_state = wait_base+2;
-            return false;
-        }
-        cur_sync->done = true;
-        // Wake up other flushers
-        bs->ringloop->wakeup();
-    }
-resume_3:
-    if (!cur_sync->done)
-    {
-        wait_state = wait_base+3;
-        return false;
-    }
-    cur_sync->done_count++;
-    if (cur_sync->done_count >= cur_sync->member_count)
-    {
-        flusher->data_syncs.erase(cur_sync);
-        cur_sync = flusher->data_syncs.end();
-    }
-    return true;
-}
-
 bool journal_flusher_co::fsync_meta(int wait_base)
 {
     if (wait_state == wait_base)        goto resume_0;
     else if (wait_state == wait_base+1) goto resume_1;
     else if (wait_state == wait_base+2) goto resume_2;
+resume_0:
     if (bs->dsk.disable_meta_fsync)
     {
         return true;
     }
-resume_0:
     if (flusher->wanting_meta_fsync || flusher->fsyncing_meta > 0)
     {
         wait_state = wait_base;
@@ -827,30 +693,69 @@ resume_2:
     return true;
 }
 
-bool journal_flusher_co::fsync_buffer(int wait_base)
+int journal_flusher_co::fsync_buffer(int wait_base)
+{
+    if (wait_state == wait_base)        goto resume_0;
+    else if (wait_state == wait_base+1) goto resume_1;
+    if (bs->dsk.disable_journal_fsync && bs->dsk.disable_meta_fsync && bs->dsk.disable_data_fsync || !bs->unsynced_big_write_count && !bs->unsynced_small_write_count)
+    {
+        return 1;
+    }
+    if (flusher->syncing_buffer)
+    {
+        return 0;
+    }
+    flusher->active_flushers++;
+    flusher->syncing_buffer++;
+resume_0:
+    assert(!wait_count);
+    compact_lsn = bs->heap->get_completed_lsn();
+    if (!bs->submit_fsyncs(wait_count))
+    {
+        wait_state = wait_base+0;
+        return 0;
+    }
+resume_1:
+    if (wait_count > 0)
+    {
+        wait_state = wait_base+1;
+        return 0;
+    }
+    bs->heap->mark_lsn_fsynced(compact_lsn);
+    flusher->active_flushers--;
+    flusher->syncing_buffer--;
+    return 2;
+}
+
+bool journal_flusher_co::trim_lsn(int wait_base)
 {
     if (wait_state == wait_base)        goto resume_0;
     else if (wait_state == wait_base+1) goto resume_1;
     else if (wait_state == wait_base+2) goto resume_2;
-    if (!bs->has_unsynced())
+    else if (wait_state == wait_base+3) goto resume_3;
+    else if (wait_state == wait_base+4) goto resume_4;
+    compact_lsn = bs->heap->get_compacted_lsn();
+    if (((blockstore_meta_header_v3_t*)bs->meta_superblock)->compacted_lsn == compact_lsn)
     {
         return true;
     }
-resume_0:
-    if (flusher->syncing_buffer)
-    {
-        wait_state = wait_base+0;
-        return false;
-    }
     flusher->active_flushers++;
-    flusher->syncing_buffer++;
-resume_1:
     assert(!wait_count);
-    fsynced_lsn = bs->heap->get_completed_lsn();
-    if (!bs->submit_fsyncs(wait_count))
+    if (!bs->dsk.disable_meta_fsync)
     {
-        wait_state = wait_base+1;
-        return false;
+        await_sqe(0);
+        data->iov = { 0 };
+        data->callback = simple_callback_w;
+        io_uring_prep_fsync(sqe, bs->dsk.meta_fd, IORING_FSYNC_DATASYNC);
+        wait_count++;
+    }
+    if (!bs->dsk.disable_data_fsync && bs->dsk.data_fd != bs->dsk.meta_fd)
+    {
+        await_sqe(1);
+        data->iov = { 0 };
+        data->callback = simple_callback_w;
+        io_uring_prep_fsync(sqe, bs->dsk.data_fd, IORING_FSYNC_DATASYNC);
+        wait_count++;
     }
 resume_2:
     if (wait_count > 0)
@@ -858,37 +763,23 @@ resume_2:
         wait_state = wait_base+2;
         return false;
     }
-    bs->heap->mark_lsn_fsynced(fsynced_lsn);
-    flusher->active_flushers--;
-    flusher->syncing_buffer--;
-    return true;
-}
-
-bool journal_flusher_co::trim_lsn(int wait_base)
-{
-    if (wait_state == wait_base)        goto resume_0;
-    else if (wait_state == wait_base+1) goto resume_1;
-    fsynced_lsn = bs->heap->get_fsynced_lsn();
-    if (((blockstore_meta_header_v3_t*)bs->meta_superblock)->completed_lsn == fsynced_lsn)
-    {
-        return true;
-    }
-    flusher->active_flushers++;
-    ((blockstore_meta_header_v3_t*)bs->meta_superblock)->completed_lsn = fsynced_lsn;
+    ((blockstore_meta_header_v3_t*)bs->meta_superblock)->compacted_lsn = compact_lsn;
     ((blockstore_meta_header_v3_t*)bs->meta_superblock)->set_crc32c();
-    await_sqe(0);
+    await_sqe(3);
     data->iov = (struct iovec){ bs->meta_superblock, (size_t)bs->dsk.meta_block_size };
     data->callback = simple_callback_w;
     io_uring_prep_writev(sqe, bs->dsk.meta_fd, &data->iov, 1, bs->dsk.meta_offset);
     // Update superblock with datasync
     sqe->rw_flags = RWF_DSYNC;
     wait_count++;
-resume_1:
+resume_4:
     if (wait_count > 0)
     {
-        wait_state = wait_base+1;
+        wait_state = wait_base+4;
         return false;
     }
+    bs->heap->mark_lsn_trimmed(compact_lsn);
+    flusher->compact_counter++;
     flusher->active_flushers--;
     return true;
 }

src/blockstore/blockstore_flush.h

@@ -6,7 +6,7 @@ struct copy_buffer_t
     uint32_t copy_flags;
     uint64_t offset, len, disk_loc, disk_offset, disk_len;
     uint8_t *buf;
-    heap_entry_t *wr;
+    uint64_t wr_lsn;
 };
 
 struct meta_sector_t
@@ -25,15 +25,6 @@ struct flusher_meta_write_t
     std::map<uint64_t, meta_sector_t>::iterator it;
 };
 
-struct flusher_data_sync_t
-{
-    int member_count = 0;
-    int ready_count = 0;
-    int done_count = 0;
-    bool sent = false;
-    bool done = false;
-};
-
 class journal_flusher_t;
 
 // Journal flusher coroutine
@@ -45,29 +36,27 @@ class journal_flusher_co
     int wait_state, wait_count;
     struct io_uring_sqe *sqe;
     struct ring_data_t *data;
-    uint8_t *new_csums = NULL;
-    uint8_t *new_bmp = NULL;
-    uint8_t *punch_bmp = NULL;
-    uint8_t *new_ext_bmp = NULL;
 
     std::function<void(ring_data_t*)> simple_callback_r, simple_callback_w;
 
     object_id cur_oid;
-    heap_entry_t *cur_obj;
-    uint64_t fsynced_lsn;
-    heap_compact_t compact_info;
-    uint64_t clean_loc;
+    uint64_t copy_id;
+    uint64_t compact_lsn;
+    uint64_t cur_version;
+    heap_object_t *cur_obj;
+    heap_write_t *begin_wr, *end_wr;
     uint32_t modified_block;
-    bool bitmap_copied;
     bool should_repeat;
 
     std::vector<copy_buffer_t> read_vec;
-    std::vector<heap_entry_t*> csum_copy;
     uint32_t overwrite_start, overwrite_end;
+    uint32_t big_start, big_end;
     int i, res;
     bool read_to_fill_incomplete;
     int copy_count;
-    std::list<flusher_data_sync_t>::iterator cur_sync;
+    uint64_t clean_loc;
+    flusher_meta_write_t meta_old, meta_new;
+    bool do_repeat = false;
 
     friend class journal_flusher_t;
 
@@ -78,10 +67,8 @@ class journal_flusher_co
     bool calc_block_checksums();
     bool write_meta_block(int wait_base);
     bool read_buffered(int wait_base);
-    void init_fsync_data();
-    bool fsync_data(int wait_base);
     bool fsync_meta(int wait_base);
-    bool fsync_buffer(int wait_base);
+    int fsync_buffer(int wait_base);
     bool trim_lsn(int wait_base);
 public:
     journal_flusher_co();
@@ -98,9 +85,10 @@ class journal_flusher_t
     blockstore_impl_t *bs;
     friend class journal_flusher_co;
 
-    robin_hood::unordered_flat_set<object_id> flushing;
+    int advance_lsn_counter = 0;
+    uint64_t compact_counter = 0;
+
     int active_flushers = 0;
-    std::list<flusher_data_sync_t> data_syncs;
     int wanting_meta_fsync = 0;
     bool fsyncing_meta = false;
     int syncing_buffer = 0;
@@ -110,6 +98,7 @@ public:
     ~journal_flusher_t();
     void loop();
     int get_syncing_buffer();
+    uint64_t get_compact_counter();
     bool is_active();
     void request_trim();
     void release_trim();

src/blockstore/blockstore_heap.h

@@ -1,4 +1,4 @@
-// Metadata storage version 3 ("lsm heap")
+// Metadata storage version 3 ("heap")
 // Copyright (c) Vitaliy Filippov, 2025+
 // License: VNPL-1.1 (see README.md for details)
 
@@ -19,332 +19,326 @@ struct pool_shard_settings_t
 {
     uint32_t pg_count;
     uint32_t pg_stripe_size;
-    uint32_t no_inode_stats;
 };
 
-#define BS_HEAP_TYPE 0x07
-#define BS_HEAP_BIG_WRITE 1
+#define BS_HEAP_TYPE 7
+#define BS_HEAP_OBJECT 1
 #define BS_HEAP_SMALL_WRITE 2
-#define BS_HEAP_INTENT_WRITE 3
-#define BS_HEAP_BIG_INTENT 4
-#define BS_HEAP_DELETE 5
-#define BS_HEAP_COMMIT 6
-#define BS_HEAP_ROLLBACK 7
-#define BS_HEAP_STABLE 0x40
-#define BS_HEAP_GARBAGE 0x80
+#define BS_HEAP_BIG_WRITE 3
+#define BS_HEAP_TOMBSTONE 4
+#define BS_HEAP_INTENT_WRITE 5
+#define BS_HEAP_STABLE 8
 
 class blockstore_heap_t;
 
-struct heap_small_write_t;
-struct heap_big_write_t;
-struct heap_big_intent_t;
-
-struct __attribute__((__packed__)) heap_entry_t
+struct __attribute__((__packed__)) heap_small_write_t
 {
     uint16_t size;
-    uint16_t entry_type;
-    uint32_t crc32c;
+    int16_t next_pos;
+    uint8_t flags;
     uint64_t lsn;
-    uint64_t inode;
-    uint64_t stripe;
     uint64_t version;
+    uint64_t location;
+    uint32_t offset;
+    uint32_t len;
+};
+
+struct __attribute__((__packed__)) heap_big_write_t
+{
+    uint16_t size;
+    int16_t next_pos;
+    uint8_t flags;
+    uint64_t lsn;
+    uint64_t version;
+    uint32_t block_num;
+};
+
+struct __attribute__((__packed__)) heap_tombstone_t
+{
+    uint16_t size;
+    int16_t next_pos;
+    uint8_t flags;
+    uint64_t lsn;
+    uint64_t version;
+};
+
+struct __attribute__((__packed__)) heap_write_t
+{
+    // size should have top bit cleared
+    uint16_t size = 0;
+    int16_t next_pos = 0;
+    uint8_t entry_type = 0; // BS_HEAP_*
+    uint64_t lsn = 0;
+    uint64_t version = 0;
 
     // uint8_t[] external_bitmap
     // uint8_t[] internal_bitmap
     // uint32_t[] checksums
 
+    heap_write_t *next();
     inline uint8_t type() const { return (entry_type & BS_HEAP_TYPE); }
     inline heap_small_write_t& small() { return *(heap_small_write_t*)this; }
     inline heap_big_write_t& big() { return *(heap_big_write_t*)this; }
-    inline heap_big_intent_t& big_intent() { return *(heap_big_intent_t*)this; }
-    bool is_garbage() const;
-    void set_garbage();
-    bool is_overwrite() const;
-    bool is_compactable() const;
-    bool is_before(const heap_entry_t *other) const;
     uint32_t get_size(blockstore_heap_t *heap);
+    uint32_t get_csum_size(blockstore_heap_t *heap);
+    bool needs_recheck(blockstore_heap_t *heap);
+    bool needs_compact(blockstore_heap_t *heap);
+    bool is_compacted(uint64_t compacted_lsn);
+    bool can_be_collapsed(blockstore_heap_t *heap);
+    bool is_allowed_before_compacted(uint64_t compacted_lsn, bool is_last_entry);
     uint8_t *get_ext_bitmap(blockstore_heap_t *heap);
     uint8_t *get_int_bitmap(blockstore_heap_t *heap);
     uint8_t *get_checksums(blockstore_heap_t *heap);
     uint32_t *get_checksum(blockstore_heap_t *heap);
     uint64_t big_location(blockstore_heap_t *heap);
     void set_big_location(blockstore_heap_t *heap, uint64_t location);
+};
+
+struct __attribute__((__packed__)) heap_object_t
+{
+    // size should have top bit cleared
+    uint16_t size = 0;
+    // linked list of write entries...
+    // newest entries are stored first to simplify scanning
+    int16_t write_pos = 0;
+    uint8_t entry_type = 0; // BS_HEAP_*
+    uint32_t crc32c = 0;
+    uint64_t inode = 0;
+    uint64_t stripe = 0;
+
+    heap_write_t *get_writes();
     uint32_t calc_crc32c();
 };
 
-struct __attribute__((__packed__)) heap_small_write_t
+struct heap_object_lsn_t
 {
-    heap_entry_t hdr;
-
-    uint64_t location;
-    uint32_t offset;
-    uint32_t len;
-
-    // Also includes 1 bitmap and 1 crc32c after the bitmap if checksums are disabled
+    object_id oid;
+    uint64_t lsn;
 };
 
-struct __attribute__((__packed__)) heap_big_write_t
+inline bool operator < (const heap_object_lsn_t & a, const heap_object_lsn_t & b)
 {
-    heap_entry_t hdr;
+    return a.oid < b.oid || a.oid == b.oid && a.lsn < b.lsn;
+}
 
-    uint32_t block_num;
+struct tmp_compact_item_t
+{
+    object_id oid;
+    uint64_t lsn;
+    bool compact;
 };
 
-struct __attribute__((__packed__)) heap_big_intent_t
+struct heap_mvcc_copy_id_t
 {
-    heap_entry_t hdr;
-
-    uint32_t block_num;
-    uint32_t offset;
-    uint32_t len;
-
-    // Also includes 2 bitmaps and 1 crc32c if checksums are disabled
+    object_id oid;
+    uint64_t copy_id;
 };
 
-struct __attribute__((__packed__)) heap_list_item_t
+inline bool operator == (const heap_mvcc_copy_id_t & a, const heap_mvcc_copy_id_t & b)
 {
-    heap_list_item_t *prev;
-    heap_list_item_t *next;
-    uint32_t block_num;
-    heap_entry_t entry;
+    return a.oid.inode == b.oid.inode && a.oid.stripe == b.oid.stripe && a.copy_id == b.copy_id;
+}
+
+namespace std
+{
+    template<> struct hash<heap_mvcc_copy_id_t>
+    {
+        inline size_t operator()(const heap_mvcc_copy_id_t &s) const
+        {
+            size_t seed = std::hash<object_id>()(s.oid);
+            // Copy-pasted from spp::hash_combine()
+            seed ^= (s.copy_id + 0xc6a4a7935bd1e995 + (seed << 6) + (seed >> 2));
+            return seed;
+        }
+    };
 };
 
 struct heap_object_mvcc_t
 {
     uint32_t readers = 0;
-    heap_entry_t *garbage_entry = NULL;
+    heap_object_t *entry_copy = NULL;
 };
 
-struct heap_block_info_t
+struct __attribute__((__packed__)) heap_block_info_t
 {
-    struct __attribute__((__packed__))
-    {
-        uint32_t used_space = 0;
-        uint32_t garbage_space = 0;
-    };
-    uint64_t mod_lsn = 0, mod_lsn_to = 0; // only 1 block write of LSN sequence is allowed at a moment
-    bool is_writing = false;
-    std::vector<heap_list_item_t*> entries;
+    uint32_t used_space = 0;
+    uint32_t free_pos = 0;
+    uint8_t *data = NULL;
 };
 
 struct heap_inflight_lsn_t
 {
+    object_id oid;
     uint64_t flags;
-    heap_entry_t *wr;
 };
 
-struct heap_compact_t
+struct heap_refqi_t
 {
-    uint64_t compact_lsn, compact_version;
-    heap_entry_t *clean_wr;
-    bool do_delete;
-};
-
-struct heap_reshard_state_t;
-
-struct heap_li_hash
-{
-    size_t operator()(const heap_list_item_t* li) const noexcept
-    {
-        return robin_hood::hash_int(li->entry.stripe);
-    }
-};
-
-struct heap_li_equal
-{
-    constexpr bool operator()(const heap_list_item_t* a, const heap_list_item_t* b) const noexcept
-    {
-        return a->entry.stripe == b->entry.stripe;
-    }
-};
-
-struct heap_recheck_state_t
-{
-    heap_entry_t *obj = NULL;
-    heap_entry_t *next_wr = NULL;
-    size_t total_reads = 0;
-    size_t sent_reads = 0;
-    size_t checked_reads = 0;
-    heap_entry_t *bad_wr = NULL;
+    uint64_t lsn;
+    uint64_t inode;
+    uint64_t location;
+    uint32_t len;
+    bool is_data;
 };
 
 using i64hash_t = robin_hood::hash<uint64_t>;
-using heap_inode_map_t = robin_hood::unordered_flat_set<heap_list_item_t*, heap_li_hash, heap_li_equal, 88>;
 using heap_block_index_t = robin_hood::unordered_flat_map<uint64_t,
-    robin_hood::unordered_flat_map<inode_t, void*, i64hash_t>, i64hash_t>;
-using heap_mvcc_map_t = robin_hood::unordered_flat_map<object_id, heap_object_mvcc_t>;
+    robin_hood::unordered_flat_map<inode_t, robin_hood::unordered_flat_map<uint64_t, uint64_t, i64hash_t, std::equal_to<uint64_t>, 88>, i64hash_t>, i64hash_t>;
+using heap_mvcc_map_t = robin_hood::unordered_flat_map<heap_mvcc_copy_id_t, heap_object_mvcc_t>;
 
 class blockstore_heap_t
 {
-    friend struct heap_entry_t;
+    friend class heap_write_t;
+    friend class heap_object_t;
 
     blockstore_disk_t *dsk = NULL;
     uint8_t* buffer_area = NULL;
+    bool abort_on_corruption = false;
+    bool abort_on_overlap = true;
     int log_level = 0;
-    const uint32_t meta_block_count = 0;
-    const uint32_t max_entry_size = 0;
 
+    const uint32_t meta_block_count = 0;
+    uint32_t target_block_free_space = 800;
+
+    uint64_t next_lsn = 0;
     robin_hood::unordered_flat_map<pool_id_t, pool_shard_settings_t> pool_shard_settings;
     // PG => inode => stripe => block number
     heap_block_index_t block_index;
     std::vector<heap_block_info_t> block_info;
     allocator_t *data_alloc = NULL;
     multilist_index_t *meta_alloc = NULL;
-    uint32_t meta_nearfull_blocks = 0;
+    uint32_t meta_alloc_count = 0;
     uint64_t meta_used_space = 0;
     multilist_alloc_t *buffer_alloc = NULL;
+    heap_mvcc_map_t object_mvcc;
+    std::unordered_map<uint64_t, uint32_t> mvcc_data_refs;
+    std::unordered_map<uint64_t, uint32_t> mvcc_buffer_refs;
     std::map<uint64_t, uint64_t> inode_space_stats;
     uint64_t buffer_area_used_space = 0;
     uint64_t data_used_space = 0;
 
-    uint64_t live_entries = 0;
-    uint64_t live_memory = 0;
-    uint64_t garbage_entries = 0;
-    uint64_t garbage_memory = 0;
-
-    uint64_t next_lsn = 0;
-    uint32_t last_allocated_block = UINT32_MAX;
-    heap_mvcc_map_t object_mvcc;
-
-    // LSN queue: inflight (writing) -> completed [-> fsynced]
+    // LSN queue: inflight (writing) -> completed [-> fsynced] -> compactable -> compacted [-> fsynced] -> trimmed and removed
     std::deque<heap_inflight_lsn_t> inflight_lsn;
     uint32_t to_compact_count = 0;
-    uint64_t compacted_count = 0;
-    uint32_t inflight_overwrite_count = 0;
     uint64_t first_inflight_lsn = 0;
     uint64_t completed_lsn = 0;
     uint64_t fsynced_lsn = 0;
-    std::deque<object_id> compact_queue;
+    uint64_t compacted_lsn = 0;
+    uint64_t next_compact_lsn = 0;
+    std::deque<heap_refqi_t> overwrite_ref_queue;
 
-    bool marked_used_blocks = false;
-    bool recheck_queue_filled = false;
-    std::vector<heap_list_item_t*> postponed_items;
-    std::set<uint32_t> recheck_modified_blocks;
-    std::deque<heap_entry_t*> recheck_queue;
-    std::map<heap_entry_t*, heap_recheck_state_t> recheck_states;
-    size_t recheck_pending_reads = 0;
+    std::vector<tmp_compact_item_t> tmp_compact_queue;
+    std::deque<object_id> recheck_queue;
     int recheck_in_progress = 0;
     bool in_recheck = false;
     std::function<void(bool is_data, uint64_t offset, uint64_t len, uint8_t* buf, std::function<void()>)> recheck_cb;
     int recheck_queue_depth = 0;
 
-    uint64_t get_pg_id(inode_t inode, uint64_t stripe);
-    bool validate_object(heap_entry_t *obj);
-    void fill_recheck_queue();
-    void recheck_drop_entries(heap_entry_t *obj, heap_entry_t *bad_wr);
-    void recheck_start_reads(heap_recheck_state_t *st);
-    int mark_used_blocks();
-    void init_free_bad_entry(heap_entry_t *wr);
-    void init_erase_bad_entry(heap_list_item_t *li);
-    bool init_erase_double_claim(heap_list_item_t *prev_li, heap_list_item_t *cur_li);
-    void recheck_full_gc();
-    void recheck_buffer(heap_entry_t *cwr, uint8_t *buf);
-    void defragment_block(uint32_t block_num);
-    void reshard_add(heap_reshard_state_t *st, heap_list_item_t *li);
+    const uint32_t max_write_entry_size;
+
+    uint64_t get_pg_id(inode_t inode, uint64_t stripe);
+    void defragment_block(uint32_t block_num);
+    uint32_t find_block_run(heap_block_info_t & block, uint32_t space);
+    uint32_t find_block_space(uint32_t block_num, uint32_t space);
+    uint32_t block_has_compactable(uint8_t *data);
+    uint32_t compact_object_to(heap_object_t *obj, uint64_t lsn, uint8_t *new_csums, bool do_free);
+    void copy_full_object(uint8_t *dst, heap_object_t *obj);
+    bool mvcc_save_copy(heap_object_t *obj);
+    bool mvcc_check_tracking(object_id oid);
+    void free_mvcc(heap_mvcc_map_t::iterator mvcc_it);
+    void allocate_block(heap_block_info_t & inf);
+    int allocate_new_object(object_id oid, uint32_t full_object_size, uint32_t *modified_block, heap_object_t **new_obj);
+    int add_object(object_id oid, heap_write_t *wr, uint32_t *modified_block);
+    void mark_overwritten(uint64_t over_lsn, uint64_t inode, heap_write_t *wr, heap_write_t *end_wr, bool tracking_active);
+    int update_object(uint32_t block_num, heap_object_t *obj, heap_write_t *wr, uint32_t *modified_block, uint32_t *moved_from_block);
+    void init_erase(uint32_t block_num, heap_object_t *obj);
+    void erase_object(uint32_t block_num, heap_object_t *obj, uint64_t lsn, bool tracking_active);
+    void reindex_block(uint32_t block_num, heap_object_t *from_obj);
+    void erase_block_index(inode_t inode, uint64_t stripe);
+    void deref_data(uint64_t inode, uint64_t location, bool free_at_0);
+    void deref_buffer(uint64_t inode, uint64_t location, uint32_t len, bool free_at_0);
+    void deref_overwrites(uint64_t lsn);
+    void free_object_space(inode_t inode, heap_write_t *from, heap_write_t *to, int mode = 0);
+    void add_used_space(uint32_t block_num, int32_t used_delta);
+    void push_inflight_lsn(object_id oid, uint64_t lsn, uint64_t flags);
 
-    void gc_block(heap_block_info_t & inf);
-    int allocate_entry(uint32_t entry_size, uint32_t *block_num, bool allow_last_free);
-    void insert_list_items(heap_list_item_t** v, size_t count, bool postpone);
-    void remove_list_item(heap_list_item_t *li);
-    void unlink_list_item(heap_list_item_t *li);
-    int add_entry(uint32_t wr_size, uint32_t *modified_block, bool allow_last_free,
-        bool explicit_complete, std::function<void(heap_entry_t *wr)> fill_entry);
-    int add_simple(heap_entry_t *obj, uint64_t version, uint32_t *modified_block, uint32_t entry_type);
-    uint32_t meta_alloc_pos(const heap_block_info_t & inf);
-    void modify_alloc(uint32_t block_num, std::function<void(heap_block_info_t &)> change_cb);
-    void mark_garbage_up_to(heap_entry_t *wr);
-    void mark_garbage(uint32_t block_num, heap_entry_t *prev_wr, uint32_t used_big);
-    void push_inflight_lsn(uint64_t lsn, heap_entry_t *wr, uint64_t flags);
-    void mark_completed_lsns(uint64_t mod_lsn);
-    void apply_inflight(heap_inflight_lsn_t & inflight);
 public:
     blockstore_heap_t(blockstore_disk_t *dsk, uint8_t *buffer_area, int log_level = 0);
     ~blockstore_heap_t();
-    void start_load(uint64_t completed_lsn);
-    // load data from the disk, returns EDOM on corruption
-    int read_blocks(uint64_t disk_offset, uint64_t size, uint8_t *buf, bool allow_corrupted,
-        std::function<void(uint32_t block_num, heap_entry_t* wr)> handle_write,
-        std::function<void(uint32_t, uint32_t, uint8_t*)> handle_block);
-    int load_blocks(uint64_t disk_offset, uint64_t size, uint8_t *buf,
-        bool allow_corrupted, uint64_t &entries_loaded);
-    // finish loading - should be called after load_blocks
+    // set initially compacted lsn - should be done before loading
+    void set_compacted_lsn(uint64_t compacted_lsn);
+    uint64_t get_compacted_lsn();
+    // load data from the disk, returns count of loaded write entries
+    void read_blocks(uint64_t disk_offset, uint64_t size, uint8_t *buf,
+        std::function<void(heap_object_t*)> handle_object, std::function<void(uint32_t, uint32_t, uint8_t*)> handle_block);
+    uint64_t load_blocks(uint64_t disk_offset, uint64_t size, uint8_t *buf);
+    // finish loading
     void finish_load();
-    // get blocks which are modified during loading and should be written to the disk
-    // before finishing initialization if not R/O
-    std::vector<uint32_t> get_recheck_modified_blocks();
     // recheck small write data after reading the database from disk
     bool recheck_small_writes(std::function<void(bool is_data, uint64_t offset, uint64_t len, uint8_t* buf, std::function<void()>)> read_buffer, int queue_depth);
-    int finish_recheck();
+    // initialize metadata area (fill it with empty data)
+    // returns 0 when done, EAGAIN when the caller has to wait more
+    int initialize();
+    // read from the metadata area
+    // returns 0 when done, EAGAIN when the caller has to wait more
+    int read();
     // reshard database according to the pool's PG count
-    void* reshard_start(pool_id_t pool, uint32_t pg_count, uint32_t pg_stripe_size, uint64_t chunk_limit);
-    bool reshard_continue(void* reshard_state, uint64_t chunk_limit);
-    bool reshard_check(pool_id_t pool, uint32_t pg_count, uint32_t pg_stripe_size);
-    void set_no_inode_stats(const std::vector<uint64_t> & pool_ids);
-    void recalc_inode_space_stats(uint64_t pool_id, bool per_inode);
+    void reshard(pool_id_t pool, uint32_t pg_count, uint32_t pg_stripe_size);
     // read an object entry and lock it against removal
     // in the future, may become asynchronous
-    heap_entry_t *lock_and_read_entry(object_id oid);
+    heap_object_t *lock_and_read_entry(object_id oid, uint64_t & copy_id);
+    // re-read a locked object entry with the given lsn (pointer may be invalidated)
+    heap_object_t *read_locked_entry(object_id oid, uint64_t copy_id);
     // read an object entry without locking it
-    heap_entry_t *read_entry(object_id oid);
+    heap_object_t *read_entry(object_id oid, uint32_t *block_num_ptr, bool for_update = false);
     // unlock an entry
-    bool unlock_entry(object_id oid);
+    bool unlock_entry(object_id oid, uint64_t copy_id);
     // set or verify checksums in a write request
-    bool calc_checksums(heap_entry_t *wr, uint8_t *data, bool set, uint32_t offset = UINT32_MAX, uint32_t len = UINT32_MAX);
+    bool calc_checksums(heap_write_t *wr, uint8_t *data, bool set, uint32_t offset = 0, uint32_t len = 0);
     // set or verify raw block checksums
     bool calc_block_checksums(uint32_t *block_csums, uint8_t *data, uint8_t *bitmap, uint32_t start, uint32_t end,
         bool set, std::function<void(uint32_t, uint32_t, uint32_t)> bad_block_cb);
     bool calc_block_checksums(uint32_t *block_csums, uint8_t *bitmap,
         uint32_t start, uint32_t end, std::function<uint8_t*(uint32_t start, uint32_t & len)> next,
         bool set, std::function<void(uint32_t, uint32_t, uint32_t)> bad_block_cb);
-    // adds a small_write or intent_write entry to an object
+    // copy an object as is
+    int copy_object(heap_object_t *obj, uint32_t *modified_block);
+    // auto-compacts the object, then adds a write entry to it and to the compaction queue
     // return 0 if OK, or maybe ENOSPC
-    int add_small_write(object_id oid, heap_entry_t **obj_ptr, uint16_t type, uint64_t version,
-        uint32_t offset, uint32_t len, uint64_t location, uint8_t *bitmap, uint8_t *data, uint32_t *modified_block);
-    // adds a big_write (overwrite) entry to an object
-    int add_big_write(object_id oid, heap_entry_t *old_head, bool stable, uint64_t version,
-        uint32_t offset, uint32_t len, uint64_t location, uint8_t *bitmap, uint8_t *data, uint32_t *modified_block);
-    // adds a "redirecting" big_intent entry to an object (same as big_write, used to avoid fsync on desktop SSDs)
-    int add_redirect_intent(object_id oid, heap_entry_t **obj_ptr, uint64_t version,
-        uint32_t offset, uint32_t len, uint64_t location, uint8_t *bitmap, uint8_t *data, uint32_t *modified_block);
-    // adds a big_intent (atomic partial modification) entry to an object
-    int add_big_intent(object_id oid, heap_entry_t **obj_ptr, uint64_t version,
-        uint32_t offset, uint32_t len, uint8_t *bitmap, uint8_t *data, uint8_t *checksums, uint32_t *modified_block);
-    // adds a compacted up to <version> entry to an object
-    int add_compact(heap_entry_t *obj, uint64_t compact_version, uint64_t compact_lsn, uint64_t compact_location,
-        bool do_delete, uint32_t *modified_block, uint8_t *new_int_bitmap, uint8_t *new_ext_bitmap, uint8_t *new_csums);
-    // "punch holes" in a big_entry
-    int punch_holes(heap_entry_t *wr, uint8_t *new_bitmap, uint8_t *new_csums, uint32_t *modified_block);
+    int post_write(object_id oid, heap_write_t *wr, uint32_t *modified_block, uint32_t *moved_from_block);
+    int post_write(uint32_t & block_num, object_id oid, heap_object_t *obj, heap_write_t *wr, uint32_t *moved_from_block);
     // stabilize an unstable object version
     // return 0 if OK, ENOENT if not exists
-    int add_commit(heap_entry_t *obj, uint64_t version, uint32_t *modified_block);
+    int post_stabilize(object_id oid, uint64_t version, uint32_t *modified_block, uint64_t *new_lsn, uint64_t *new_to_lsn);
     // rollback an unstable object version
     // return 0 if OK, ENOENT if not exists, EBUSY if already stable
-    int add_rollback(heap_entry_t *obj, uint64_t version, uint32_t *modified_block);
+    int post_rollback(object_id oid, uint64_t version, uint64_t *new_lsn, uint32_t *modified_block);
     // forget an object
     // return error code
-    int add_delete(heap_entry_t *obj, uint32_t *modified_block);
+    int post_delete(object_id oid, uint64_t *new_lsn, uint32_t *modified_block);
+    int post_delete(uint32_t block_num, heap_object_t *obj, uint64_t *new_lsn);
     // get the next object to compact
     // guaranteed to return objects in min lsn order
     // returns 0 if OK, ENOENT if nothing to compact
     int get_next_compact(object_id & oid);
-    void iterate_with_stable(heap_entry_t *obj, uint64_t max_lsn, std::function<bool(heap_entry_t*, bool stable)> cb);
-    // iterate compactable entries
-    heap_compact_t iterate_compaction(heap_entry_t *obj, uint64_t fsynced_lsn, bool under_pressure,
-        std::function<void(heap_entry_t*)> small_wr_cb);
-    // iterate all objects
-    void iterate_objects(std::function<void(heap_entry_t*, uint32_t block_num)> cb);
+    // get the range of an object eligible for compaction
+    void get_compact_range(heap_object_t *obj, uint64_t max_lsn, heap_write_t **begin_wr, heap_write_t **end_wr);
+    // mark an object as compacted up to the given lsn
+    int compact_object(object_id oid, uint64_t lsn, uint8_t *new_csums);
     // retrieve object listing from a PG
     int list_objects(uint32_t pg_num, object_id min_oid, object_id max_oid,
         obj_ver_id **result_list, size_t *stable_count, size_t *unstable_count);
+    // set a block number for a new object and returns error status: 0, EAGAIN or ENOSPC
+    int get_block_for_new_object(uint32_t & out_block_num, uint32_t size = 0);
 
     // inflight write tracking
-    void start_block_write(uint32_t block_num);
-    void complete_block_write(uint32_t block_num);
-    void complete_lsn_write(uint64_t lsn);
-    bool is_lsn_completed(uint64_t lsn);
+    void mark_lsn_completed(uint64_t lsn);
+    void mark_lsn_fsynced(uint64_t lsn);
+    void mark_lsn_compacted(uint64_t lsn, bool allow_undone = false);
+    void mark_object_compacted(heap_object_t *obj, uint64_t max_lsn);
+    void mark_lsn_trimmed(uint64_t lsn);
     uint64_t get_completed_lsn();
     uint64_t get_fsynced_lsn();
-    void mark_lsn_fsynced(uint64_t lsn);
 
     // data device block allocator functions
     uint64_t find_free_data();
@@ -360,8 +354,7 @@ public:
     uint64_t get_buffer_area_used_space();
 
     // get metadata block data buffer and used space
-    void get_meta_block(uint32_t block_num, uint8_t *buffer);
-    void fill_block_empty_space(uint8_t *buffer, uint32_t pos);
+    uint8_t *get_meta_block(uint32_t block_num);
     uint32_t get_meta_block_used_space(uint32_t block_num);
 
     // get space usage statistics
@@ -370,21 +363,14 @@ public:
     uint64_t get_meta_total_space();
     uint64_t get_meta_used_space();
     uint32_t get_meta_nearfull_blocks();
+    uint32_t get_inflight_queue_size();
     uint32_t get_compact_queue_size();
     uint32_t get_to_compact_count();
-    uint64_t get_compacted_count();
-    uint64_t get_live_entries();
-    uint64_t get_live_memory();
-    uint64_t get_garbage_entries();
-    uint64_t get_garbage_memory();
 
-    uint64_t entry_pos(uint32_t block_num, uint32_t offset);
-    heap_entry_t *entry_from_pos(uint64_t entry_pos, bool allow_unallocated = false);
-    heap_entry_t *prev(heap_entry_t *wr);
-    uint32_t get_simple_entry_size();
-    uint32_t get_big_entry_size();
-    uint32_t get_big_intent_entry_size();
-    uint32_t get_small_entry_size(uint32_t offset, uint32_t len);
-    uint32_t get_csum_size(heap_entry_t *wr);
-    uint32_t get_csum_size(uint32_t entry_type, uint32_t offset = 0, uint32_t len = 0);
+    // get maximum size for a temporary heap_write_t buffer
+    uint32_t get_max_write_entry_size();
+
+    // only for tests
+    void set_abort_on_corruption(bool fail);
+    void set_abort_on_overlap(bool fail);
 };

src/blockstore/blockstore_impl.cpp

@@ -23,22 +23,26 @@ blockstore_impl_t::blockstore_impl_t(blockstore_config_t & config, ring_loop_i *
         dsk.open_meta();
         dsk.open_journal();
         dsk.calc_lengths();
-        dsk.check_lengths();
+        zero_object = (uint8_t*)memalign_or_die(MEM_ALIGNMENT, dsk.data_block_size);
     }
     catch (std::exception & e)
     {
         dsk.close_all();
         throw;
     }
+    memset(zero_object, 0, dsk.data_block_size);
     meta_superblock = (uint8_t*)memalign_or_die(MEM_ALIGNMENT, dsk.meta_block_size);
     memset(meta_superblock, 0, dsk.meta_block_size);
+}
+
+void blockstore_impl_t::init()
+{
     flusher = new journal_flusher_t(this);
     if (dsk.inmemory_journal)
     {
         buffer_area = (uint8_t*)memalign_or_die(MEM_ALIGNMENT, dsk.journal_len);
     }
     heap = new blockstore_heap_t(&dsk, buffer_area, log_level);
-    ringloop->wakeup();
 }
 
 blockstore_impl_t::~blockstore_impl_t()
@@ -51,6 +55,8 @@ blockstore_impl_t::~blockstore_impl_t()
         free(buffer_area);
     if (meta_superblock)
         free(meta_superblock);
+    if (zero_object)
+        free(zero_object);
     ringloop->unregister_consumer(&ring_consumer);
     dsk.close_all();
 }
@@ -101,7 +107,6 @@ void blockstore_impl_t::loop()
         unsigned initial_ring_space = ringloop->space_left();
         int op_idx = 0, new_idx = 0;
         bool has_unfinished_writes = false;
-        bool has_unfinished_sync = false;
         for (; op_idx < submit_queue.size(); op_idx++, new_idx++)
         {
             auto op = submit_queue[op_idx];
@@ -139,13 +144,7 @@ void blockstore_impl_t::loop()
             else if (op->opcode == BS_OP_SYNC)
             {
                 // syncs only completed writes, so doesn't have to be blocked by anything
-                if (!has_unfinished_sync)
-                {
-                    wr_st = continue_sync(op);
-                    has_unfinished_sync = (wr_st != 2);
-                }
-                else
-                    wr_st = 0;
+                wr_st = continue_sync(op);
             }
             else if (op->opcode == BS_OP_STABLE || op->opcode == BS_OP_ROLLBACK)
             {
@@ -161,7 +160,9 @@ void blockstore_impl_t::loop()
                     wr_st = 2;
                 }
                 else
+                {
                     wr_st = 0;
+                }
             }
             if (wr_st == 2)
             {
@@ -191,14 +192,6 @@ void blockstore_impl_t::loop()
         {
             flusher->loop();
         }
-        for (auto & block_num: pending_modified_blocks)
-        {
-            auto & mb = modified_blocks[block_num];
-            heap->get_meta_block(block_num, mb.buf);
-            heap->start_block_write(block_num);
-            mb.sent = true;
-        }
-        pending_modified_blocks.clear();
         int ret = ringloop->submit();
         if (ret < 0)
         {
@@ -221,7 +214,7 @@ bool blockstore_impl_t::is_safe_to_stop()
     {
         return false;
     }
-    if (has_unsynced())
+    if (unsynced_big_write_count > 0 || unsynced_small_write_count > 0)
     {
         if (!readonly && !stop_sync_submitted)
         {
@@ -257,7 +250,7 @@ void blockstore_impl_t::check_wait(blockstore_op_t *op)
     }
     else if (PRIV(op)->wait_for == WAIT_COMPACTION)
     {
-        if (heap->get_compacted_count() <= PRIV(op)->wait_detail)
+        if (flusher->get_compact_counter() <= PRIV(op)->wait_detail)
         {
             // do not submit
 #ifdef BLOCKSTORE_DEBUG
@@ -280,8 +273,7 @@ void blockstore_impl_t::enqueue_op(blockstore_op_t *op)
         ((op->opcode == BS_OP_READ || op->opcode == BS_OP_WRITE || op->opcode == BS_OP_WRITE_STABLE) && (
             op->offset >= dsk.data_block_size ||
             op->len > dsk.data_block_size-op->offset ||
-            (op->offset % dsk.bitmap_granularity) ||
-            (op->len % dsk.bitmap_granularity)
+            (op->len % dsk.disk_alignment)
         )) ||
         readonly && op->opcode != BS_OP_READ && op->opcode != BS_OP_LIST)
     {
@@ -328,13 +320,9 @@ void blockstore_impl_t::process_list(blockstore_op_t *op)
         FINISH_OP(op);
         return;
     }
-    // Check if the DB is sharded correctly
-    if (!heap->reshard_check(INODE_POOL(min_inode), pg_count, pg_stripe_size))
-    {
-        op->retval = -EAGAIN;
-        FINISH_OP(op);
-        return;
-    }
+    // Check if the DB needs resharding
+    // (we don't know about PGs from the beginning, we only create "shards" here)
+    heap->reshard(INODE_POOL(min_inode), pg_count, pg_stripe_size);
     obj_ver_id *result = NULL;
     size_t stable_count = 0, unstable_count = 0;
     int res = heap->list_objects(list_pg, op->min_oid, op->max_oid, &result, &stable_count, &unstable_count);
@@ -357,7 +345,6 @@ void blockstore_impl_t::process_list(blockstore_op_t *op)
 
 void blockstore_impl_t::set_no_inode_stats(const std::vector<uint64_t> & pool_ids)
 {
-    heap->set_no_inode_stats(pool_ids);
 }
 
 void blockstore_impl_t::dump_diagnostics()
@@ -401,13 +388,3 @@ std::string blockstore_impl_t::get_op_diag(blockstore_op_t *op)
         snprintf(buf, sizeof(buf), "state=%d", priv->op_state);
     return std::string(buf);
 }
-
-void* blockstore_impl_t::reshard_start(pool_id_t pool, uint32_t pg_count, uint32_t pg_stripe_size, uint64_t chunk_limit)
-{
-    return heap->reshard_start(pool, pg_count, pg_stripe_size, chunk_limit);
-}
-
-bool blockstore_impl_t::reshard_continue(void *reshard_state, uint64_t chunk_limit)
-{
-    return heap->reshard_continue(reshard_state, chunk_limit);
-}

src/blockstore/blockstore_impl.h

@@ -29,6 +29,18 @@ class blockstore_impl_t;
 
 //#define BLOCKSTORE_DEBUG
 
+// - Sync must be submitted after previous writes/deletes (not before!)
+// - Reads may be submitted in parallel with writes/deletes because we use MVCC
+// - Writes may be submitted in any order, because they don't overlap. Each write
+//   goes into a new location - either on the journal device or on the data device
+// - Stable (stabilize) must be submitted after sync of that object is completed
+//   It's even OK to return an error to the caller if that object is not synced yet
+// - compacted_lsn should be moved forward only after all versions are moved to the main storage
+// - If an operation can not be submitted because the ring is full
+//   we should stop submission of other operations. Otherwise some "scatter" reads
+//   may end up blocked for a long time.
+// Otherwise, the submission order is free.
+
 #include "blockstore_init.h"
 
 #include "blockstore_flush.h"
@@ -41,30 +53,25 @@ struct blockstore_op_private_t
     int pending_ops;
     int op_state;
 
-    // Write, sync, stabilize
-    uint32_t modified_block, modified_block2;
+    // Read, write, sync, stabilize
+    uint64_t lsn;
 
     // Read
     std::vector<copy_buffer_t> read_vec;
 
-    // Read, write
-    uint64_t lsn;
-
     // Write
     uint64_t location;
-    uint32_t write_type;
+    bool is_big;
 
     // Stabilize, rollback
     int stab_pos;
 
-    // Write
-    timespec tv_begin;
-};
+    // Stabilize
+    uint64_t to_lsn;
 
-struct bs_modified_block_t
-{
-    bool sent;
-    uint8_t *buf;
+    // Write
+    struct iovec iov_zerofill[3];
+    timespec tv_begin;
 };
 
 class blockstore_impl_t: public blockstore_i
@@ -78,7 +85,6 @@ public:
     // Suitable only for server SSDs with capacitors, requires disabled data and journal fsyncs
     int immediate_commit = IMMEDIATE_NONE;
     bool inmemory_meta = false;
-    bool skip_corrupted_meta_entries = false;
     uint32_t meta_write_recheck_parallelism = 0;
     // Maximum and minimum flusher count
     unsigned max_flusher_count = 0, min_flusher_count = 0;
@@ -109,16 +115,13 @@ public:
     uint8_t* meta_superblock = NULL;
     uint8_t *buffer_area = NULL;
     std::vector<blockstore_op_t*> submit_queue;
-    int unsynced_data_write_count = 0, unsynced_buffer_write_count = 0, unsynced_meta_write_count = 0;
+    int unsynced_big_write_count = 0, unsynced_small_write_count = 0, unsynced_meta_write_count = 0;
     int unsynced_queued_ops = 0;
-
-    std::vector<uint32_t> pending_modified_blocks;
-    robin_hood::unordered_flat_map<uint32_t, bs_modified_block_t> modified_blocks;
+    uint8_t *zero_object = NULL;
 
     journal_flusher_t *flusher;
     int write_iodepth = 0;
     int inflight_big = 0;
-    int intent_write_counter = 0;
     bool fsyncing_data = false;
 
     bool live = false, queue_stall = false;
@@ -143,17 +146,18 @@ public:
     int metadata_buf_size;
     blockstore_init_meta* metadata_init_reader;
 
+    void init();
     void check_wait(blockstore_op_t *op);
     void init_op(blockstore_op_t *op);
 
     // Read
     int dequeue_read(blockstore_op_t *op);
     int fulfill_read(blockstore_op_t *op);
-    uint32_t prepare_read(std::vector<copy_buffer_t> & read_vec, heap_entry_t *obj, heap_entry_t *wr, uint32_t start, uint32_t end, uint32_t skip_csum);
-    uint32_t prepare_read_with_bitmaps(std::vector<copy_buffer_t> & read_vec, heap_entry_t *obj, heap_entry_t *wr, uint32_t start, uint32_t end, uint32_t skip_csum);
+    uint32_t prepare_read(std::vector<copy_buffer_t> & read_vec, heap_object_t *obj, heap_write_t *wr, uint32_t start, uint32_t end);
+    uint32_t prepare_read_with_bitmaps(std::vector<copy_buffer_t> & read_vec, heap_object_t *obj, heap_write_t *wr, uint32_t start, uint32_t end);
     uint32_t prepare_read_zero(std::vector<copy_buffer_t> & read_vec, uint32_t start, uint32_t end);
-    uint32_t prepare_read_simple(std::vector<copy_buffer_t> & read_vec, heap_entry_t *obj, heap_entry_t *wr, uint32_t start, uint32_t end, uint32_t skip_csum);
-    void prepare_disk_read(std::vector<copy_buffer_t> & read_vec, int pos, heap_entry_t *obj, heap_entry_t *wr,
+    uint32_t prepare_read_simple(std::vector<copy_buffer_t> & read_vec, heap_object_t *obj, heap_write_t *wr, uint32_t start, uint32_t end);
+    void prepare_disk_read(std::vector<copy_buffer_t> & read_vec, int pos, heap_object_t *obj, heap_write_t *wr,
         uint32_t blk_start, uint32_t blk_end, uint32_t start, uint32_t end, uint32_t copy_flags);
     void find_holes(std::vector<copy_buffer_t> & read_vec, uint32_t item_start, uint32_t item_end,
         std::function<void(int&, uint32_t, uint32_t)> callback);
@@ -163,10 +167,9 @@ public:
 
     // Write
     bool enqueue_write(blockstore_op_t *op);
-    void prepare_meta_block_write(uint32_t modified_block);
-    bool meta_block_is_pending(uint32_t modified_block);
-    bool intent_write_allowed(blockstore_op_t *op, heap_entry_t *obj);
+    void prepare_meta_block_write(blockstore_op_t *op, uint64_t modified_block, io_uring_sqe *sqe = NULL);
     int dequeue_write(blockstore_op_t *op);
+    int make_big_write(blockstore_op_t *op, uint32_t offset, uint32_t len, uint32_t *modified_block, uint32_t *moved_from_block);
     int continue_write(blockstore_op_t *op);
     void handle_write_event(ring_data_t *data, blockstore_op_t *op);
 
@@ -174,7 +177,6 @@ public:
     int continue_sync(blockstore_op_t *op);
     bool submit_fsyncs(int & wait_count);
     int do_sync(blockstore_op_t *op, int base_state);
-    bool has_unsynced();
 
     // Stabilize
     int dequeue_stable(blockstore_op_t *op);
@@ -190,9 +192,6 @@ public:
     void parse_config(blockstore_config_t & config);
     void parse_config(blockstore_config_t & config, bool init);
 
-    void* reshard_start(pool_id_t pool, uint32_t pg_count, uint32_t pg_stripe_size, uint64_t chunk_limit);
-    bool reshard_continue(void *reshard_state, uint64_t chunk_limit);
-
     // Event loop
     void loop();
 
@@ -227,11 +226,6 @@ public:
     inline uint32_t get_block_size() { return dsk.data_block_size; }
     inline uint64_t get_block_count() { return dsk.block_count; }
     uint64_t get_free_block_count();
-    inline uint32_t get_bitmap_granularity() { return dsk.bitmap_granularity; }
+    inline uint32_t get_bitmap_granularity() { return dsk.disk_alignment; }
     inline uint64_t get_journal_size() { return dsk.journal_len; }
-
-    inline uint64_t get_live_entries() { return heap->get_live_entries(); }
-    inline uint64_t get_live_memory() { return heap->get_live_memory(); }
-    inline uint64_t get_garbage_entries() { return heap->get_garbage_entries(); }
-    inline uint64_t get_garbage_memory() { return heap->get_garbage_memory(); }
 };

src/blockstore/blockstore_init.cpp

@@ -51,8 +51,6 @@ int blockstore_init_meta::loop()
     else if (wait_state == 5) goto resume_5;
     else if (wait_state == 6) goto resume_6;
     else if (wait_state == 7) goto resume_7;
-    else if (wait_state == 8) goto resume_8;
-    else if (wait_state == 9) goto resume_9;
     metadata_buffer = memalign(MEM_ALIGNMENT, 2*bs->metadata_buf_size);
     if (!metadata_buffer)
         throw std::runtime_error("Failed to allocate metadata read buffer");
@@ -72,6 +70,7 @@ resume_1:
     }
     if (is_zero((uint64_t*)bs->meta_superblock, bs->dsk.meta_block_size))
     {
+        bs->dsk.check_lengths();
         {
             blockstore_meta_header_v3_t *hdr = (blockstore_meta_header_v3_t *)bs->meta_superblock;
             hdr->zero = 0;
@@ -80,15 +79,12 @@ resume_1:
             hdr->meta_block_size = bs->dsk.meta_block_size;
             hdr->data_block_size = bs->dsk.data_block_size;
             hdr->bitmap_granularity = bs->dsk.bitmap_granularity;
+            hdr->compacted_lsn = 0;
             if (bs->dsk.meta_format >= BLOCKSTORE_META_FORMAT_V2)
             {
                 hdr->data_csum_type = bs->dsk.data_csum_type;
                 hdr->csum_block_size = bs->dsk.csum_block_size;
             }
-            if (bs->dsk.meta_format >= BLOCKSTORE_META_FORMAT_HEAP)
-            {
-                hdr->meta_area_size = bs->dsk.meta_area_size;
-            }
             hdr->set_crc32c();
         }
         if (bs->readonly)
@@ -139,30 +135,28 @@ resume_1:
             hdr->data_block_size != bs->dsk.data_block_size ||
             hdr->bitmap_granularity != bs->dsk.bitmap_granularity ||
             hdr->data_csum_type != bs->dsk.data_csum_type ||
-            hdr->csum_block_size != bs->dsk.csum_block_size ||
-            hdr->meta_area_size != bs->dsk.meta_area_size)
+            hdr->csum_block_size != bs->dsk.csum_block_size)
         {
             printf(
                 "Configuration stored in metadata superblock"
-                " (meta_block_size=%u, data_block_size=%u, bitmap_granularity=%u, data_csum_type=%u, csum_block_size=%u, meta_area_size=%ju)"
-                " differs from OSD configuration (%ju/%ju/%u, %u/%u, %ju).\n",
+                " (meta_block_size=%u, data_block_size=%u, bitmap_granularity=%u, data_csum_type=%u, csum_block_size=%u)"
+                " differs from OSD configuration (%u/%u/%u, %u/%u).\n",
                 hdr->meta_block_size, hdr->data_block_size, hdr->bitmap_granularity,
-                hdr->data_csum_type, hdr->csum_block_size, hdr->meta_area_size,
+                hdr->data_csum_type, hdr->csum_block_size,
                 bs->dsk.meta_block_size, bs->dsk.data_block_size, bs->dsk.bitmap_granularity,
-                bs->dsk.data_csum_type, bs->dsk.csum_block_size, bs->dsk.meta_area_size
+                bs->dsk.data_csum_type, bs->dsk.csum_block_size
             );
             exit(1);
         }
-        uint32_t csum = hdr->header_csum;
-        hdr->header_csum = 0;
-        if (crc32c(0, hdr, sizeof(*hdr)) != csum)
+        if (bs->dsk.meta_format != hdr->version)
         {
-            printf("Metadata header is corrupt (checksum mismatch).\n");
-            exit(1);
+            bs->dsk.meta_format = hdr->version;
+            bs->dsk.calc_lengths();
         }
-        hdr->header_csum = csum;
+        bs->dsk.check_lengths();
     }
-    bs->heap->start_load(((blockstore_meta_header_v3_t *)bs->meta_superblock)->completed_lsn);
+    bs->init();
+    bs->heap->set_compacted_lsn(((blockstore_meta_header_v3_t *)bs->meta_superblock)->compacted_lsn);
     if (bs->dsk.inmemory_journal)
     {
         // Read buffer area
@@ -216,10 +210,8 @@ resume_4:
                     io_uring_prep_readv(sqe, bs->dsk.meta_fd, &data->iov, 1, bs->dsk.meta_offset + bufs[i].offset);
                 else
                 {
-                    // Fill metadata with empty block pattern
-                    memset(bufs[i].buf, 0, bufs[i].size);
-                    for (uint64_t o = 0; o < bufs[i].size; o += bs->dsk.meta_block_size)
-                        bs->heap->fill_block_empty_space(bufs[i].buf + o, 0);
+                    // Fill metadata with zeroes
+                    memset(data->iov.iov_base, 0, data->iov.iov_len);
                     io_uring_prep_writev(sqe, bs->dsk.meta_fd, &data->iov, 1, bs->dsk.meta_offset + bufs[i].offset);
                 }
                 bs->ringloop->submit();
@@ -232,11 +224,7 @@ resume_4:
         if (bufs[i].state == INIT_META_READ_DONE)
         {
             // Handle result
-            uint64_t loaded = 0;
-            int r = bs->heap->load_blocks(bufs[i].offset-bs->dsk.meta_block_size, bufs[i].size, bufs[i].buf, bs->skip_corrupted_meta_entries, loaded);
-            if (r != 0)
-                exit(1);
-            entries_loaded += loaded;
+            entries_loaded += bs->heap->load_blocks(bufs[i].offset-bs->dsk.meta_block_size, bufs[i].size, bufs[i].buf);
             bufs[i].state = 0;
             bs->ringloop->wakeup();
         }
@@ -247,89 +235,8 @@ resume_4:
         return 1;
     }
     // metadata read finished
-    bs->heap->finish_load();
-    printf("Metadata entries loaded: %ju, rechecking unfinished writes and garbage entries\n", entries_loaded);
-    // asynchronous recheck
-resume_6:
-    wait_state = 6;
-    bs->heap->recheck_small_writes([this](bool is_data, uint64_t offset, uint64_t len, uint8_t *buf, std::function<void()> cb)
-    {
-        if (!buf)
-        {
-            wait_state = 7;
-            bs->ringloop->wakeup();
-            return;
-        }
-        GET_SQE();
-        data->iov = (iovec){ buf, len };
-        data->callback = [offset, cb](ring_data_t *data)
-        {
-            if (data->res < 0)
-            {
-                fprintf(stderr, "Buffer area read failed at offset %ju: %d\n", offset, data->res);
-                exit(1);
-            }
-            cb();
-        };
-        io_uring_prep_readv(sqe, (is_data ? bs->dsk.data_fd : bs->dsk.journal_fd), &data->iov, 1,
-            (is_data ? bs->dsk.data_offset : bs->dsk.journal_offset) + offset);
-        bs->ringloop->submit();
-    }, bs->meta_write_recheck_parallelism);
-    return 1;
-resume_7:
-    if (bs->heap->finish_recheck() != 0)
-    {
-        exit(1);
-    }
-    recheck_mod = bs->heap->get_recheck_modified_blocks();
-    if (bs->readonly)
-    {
-        recheck_mod.clear();
-        printf("Actual metadata entries: %ju\n", bs->heap->get_live_entries());
-    }
-    else
-    {
-        printf("Actual metadata entries: %ju, clearing garbage in %zu metadata blocks\n", bs->heap->get_live_entries(), recheck_mod.size());
-    }
-    for (i = 0; i < recheck_mod.size(); i++)
-    {
-resume_8:
-        if (wait_count >= bs->meta_write_recheck_parallelism || !(sqe = bs->get_sqe()))
-        {
-            bs->ringloop->submit();
-            wait_state = 8;
-            return 1;
-        }
-        uint32_t block_num = recheck_mod[i];
-        uint64_t block_offset = bs->dsk.meta_offset + (uint64_t)(block_num+1) * bs->dsk.meta_block_size;
-        data = ((ring_data_t*)sqe->user_data);
-        uint8_t *buf = (uint8_t*)memalign_or_die(MEM_ALIGNMENT, bs->dsk.meta_block_size);
-        bs->heap->get_meta_block(block_num, buf);
-        data->iov = { buf, bs->dsk.meta_block_size };
-        data->callback = [this, buf, block_offset](ring_data_t *data)
-        {
-            wait_count--;
-            free(buf);
-            if (data->res != bs->dsk.meta_block_size)
-            {
-                throw std::runtime_error(
-                    "write metadata failed at offset " + std::to_string(block_offset) + ": " + strerror(-data->res)
-                );
-            }
-        };
-        io_uring_prep_writev(sqe, bs->dsk.meta_fd, &data->iov, 1, block_offset);
-        wait_count++;
-    }
-resume_9:
-    if (wait_count > 0)
-    {
-        bs->ringloop->submit();
-        wait_state = 9;
-        return 1;
-    }
-    free(metadata_buffer);
-    metadata_buffer = NULL;
-    if (!bs->dsk.disable_meta_fsync && !bs->readonly)
+    printf("Metadata entries loaded: %ju, used blocks: %ju / %ju\n", entries_loaded, bs->heap->get_data_used_space() / bs->dsk.data_block_size, bs->dsk.block_count);
+    if (zero_on_init && !bs->dsk.disable_meta_fsync)
     {
         GET_SQE();
         io_uring_prep_fsync(sqe, bs->dsk.meta_fd, IORING_FSYNC_DATASYNC);
@@ -345,9 +252,36 @@ resume_9:
             return 1;
         }
     }
-    printf("Loading finished. Data used: %ju / %ju bytes (%s / %s)\n",
-        bs->heap->get_data_used_space(), bs->dsk.block_count * bs->dsk.data_block_size,
-        format_size(bs->heap->get_data_used_space()).c_str(),
-        format_size(bs->dsk.block_count * bs->dsk.data_block_size).c_str());
+    // asynchronous recheck
+resume_6:
+    wait_state = 6;
+    bs->heap->recheck_small_writes([this](bool is_data, uint64_t offset, uint64_t len, uint8_t *buf, std::function<void()> cb)
+    {
+        if (!buf)
+        {
+            wait_state = 7;
+            bs->ringloop->wakeup();
+            return;
+        }
+        GET_SQE();
+        data->iov = (iovec){ buf, len };
+        data->callback = [this, offset, cb](ring_data_t *data)
+        {
+            if (data->res < 0)
+            {
+                fprintf(stderr, "Buffer area read failed at offset %ju: %d\n", offset, data->res);
+                exit(1);
+            }
+            cb();
+        };
+        io_uring_prep_readv(sqe, (is_data ? bs->dsk.data_fd : bs->dsk.journal_fd), &data->iov, 1,
+            (is_data ? bs->dsk.data_offset : bs->dsk.journal_offset) + offset);
+        bs->ringloop->submit();
+    }, bs->meta_write_recheck_parallelism);
+    return 1;
+resume_7:
+    bs->heap->finish_load();
+    free(metadata_buffer);
+    metadata_buffer = NULL;
     return 0;
 }

src/blockstore/blockstore_init.h

@@ -15,7 +15,6 @@ class blockstore_init_meta
 {
     blockstore_impl_t *bs;
     int wait_state = 0;
-    int wait_count = 0;
     bool zero_on_init = false;
     void *metadata_buffer = NULL;
     blockstore_init_meta_buf bufs[2] = {};
@@ -26,7 +25,6 @@ class blockstore_init_meta
     uint64_t next_offset = 0;
     uint64_t last_read_offset = 0;
     uint64_t entries_loaded = 0;
-    std::vector<uint32_t> recheck_mod;
     int i = 0, j = 0;
     bool handle_meta_block(uint8_t *buf, uint64_t count, uint64_t done_cnt);
     void handle_event(ring_data_t *data, int buf_num);

src/blockstore/blockstore_internal.h

@@ -1,6 +1,3 @@
-// Copyright (c) Vitaliy Filippov, 2019+
-// License: VNPL-1.1 (see README.md for details)
-
 #pragma once
 
 #define BS_SUBMIT_CHECK_SQES(n) \
@@ -51,11 +48,3 @@
 #define COPY_BUF_COALESCED  0x10
 #define COPY_BUF_PADDED     0x20
 #define COPY_BUF_SKIP_CSUM  0x40
-
-#ifndef RWF_ATOMIC
-#define RWF_ATOMIC 0x40
-#endif
-
-#ifndef RWF_DSYNC
-#define RWF_DSYNC 0x02
-#endif

src/blockstore/blockstore_open.cpp

@@ -28,7 +28,6 @@ void blockstore_impl_t::parse_config(blockstore_config_t & config, bool init)
     throttle_target_parallelism = strtoull(config["throttle_target_parallelism"].c_str(), NULL, 10);
     throttle_threshold_us = strtoull(config["throttle_threshold_us"].c_str(), NULL, 10);
     perfect_csum_update = config["perfect_csum_update"] == "true" || config["perfect_csum_update"] == "1" || config["perfect_csum_update"] == "yes";
-    skip_corrupted_meta_entries = config["skip_corrupted_meta_entries"] == "true" || config["skip_corrupted_meta_entries"] == "1" || config["skip_corrupted_meta_entries"] == "yes";
     if (config["autosync_writes"] != "")
     {
         autosync_writes = strtoull(config["autosync_writes"].c_str(), NULL, 10);
@@ -43,7 +42,7 @@ void blockstore_impl_t::parse_config(blockstore_config_t & config, bool init)
     }
     if (!journal_trim_interval)
     {
-        journal_trim_interval = 4096;
+        journal_trim_interval = 1024;
     }
     if (!flusher_start_threshold)
     {
@@ -97,10 +96,6 @@ void blockstore_impl_t::parse_config(blockstore_config_t & config, bool init)
     {
         metadata_buf_size = 4*1024*1024;
     }
-    if (metadata_buf_size % dsk.meta_block_size)
-    {
-        throw std::runtime_error("metadata_buf_size should be a multiple of meta_block_size");
-    }
     if (!meta_write_recheck_parallelism)
     {
         meta_write_recheck_parallelism = 16;

src/blockstore/blockstore_read.cpp

@@ -7,7 +7,7 @@
 
 int blockstore_impl_t::dequeue_read(blockstore_op_t *op)
 {
-    heap_entry_t *obj = heap->lock_and_read_entry(op->oid);
+    heap_object_t *obj = heap->lock_and_read_entry(op->oid, PRIV(op)->lsn);
     if (!obj)
     {
         op->version = 0;
@@ -17,44 +17,16 @@ int blockstore_impl_t::dequeue_read(blockstore_op_t *op)
     }
     uint32_t fulfilled = 0;
     PRIV(op)->pending_ops = 0;
-    PRIV(op)->lsn = 0;
     auto & rv = PRIV(op)->read_vec;
     uint64_t result_version = 0;
     bool found = false;
-    uint32_t skip_csum = 0;
-    uint32_t blk_start = op->offset, blk_end = op->offset+op->len;
-    bool need_skip = dsk.csum_block_size > dsk.bitmap_granularity && !perfect_csum_update;
-    if (need_skip)
+    for (auto wr = obj->get_writes(); wr; wr = wr->next())
     {
-        PRIV(op)->lsn = obj->lsn;
-        blk_start = op->offset - op->offset%dsk.csum_block_size;
-        blk_end = op->offset + op->len;
-        if (blk_end % dsk.csum_block_size)
-            blk_end += dsk.csum_block_size - (blk_end % dsk.csum_block_size);
-    }
-    bool need_wait = false;
-    heap->iterate_with_stable(obj, obj->lsn, [&](heap_entry_t *wr, bool stable)
-    {
-        if (wr->type() == BS_HEAP_DELETE)
+        if (op->version < wr->version)
         {
-            return false;
+            continue;
         }
-        if (!heap->is_lsn_completed(wr->lsn))
-        {
-            if (wr->type() == BS_HEAP_BIG_INTENT && wr->big_intent().offset < blk_end && wr->big_intent().offset+wr->big_intent().len > blk_start ||
-                wr->type() == BS_HEAP_INTENT_WRITE && wr->small().offset < blk_end && wr->small().offset+wr->small().len > blk_start)
-            {
-                // Wait until intent write is completed
-                need_wait = true;
-                return false;
-            }
-            else if (wr->type() == BS_HEAP_SMALL_WRITE && wr->small().offset < blk_end && wr->small().offset+wr->small().len > blk_start)
-            {
-                // Skip entry and read the previous one
-                return true;
-            }
-        }
-        if (op->version >= wr->version && !found)
+        if (!found)
         {
             found = true;
             result_version = wr->version;
@@ -63,39 +35,18 @@ int blockstore_impl_t::dequeue_read(blockstore_op_t *op)
                 memcpy(op->bitmap, wr->get_ext_bitmap(heap), dsk.clean_entry_bitmap_size);
             }
         }
-        if (op->version >= wr->version)
+        fulfilled += prepare_read(PRIV(op)->read_vec, obj, wr, op->offset, op->offset+op->len);
+        if (fulfilled == op->len ||
+            wr->type() == BS_HEAP_BIG_WRITE ||
+            wr->type() == BS_HEAP_TOMBSTONE)
         {
-            fulfilled += prepare_read(PRIV(op)->read_vec, obj, wr, op->offset, op->offset+op->len,
-                wr->type() != BS_HEAP_SMALL_WRITE ? skip_csum : 0);
-            if (fulfilled == op->len ||
-                wr->type() == BS_HEAP_BIG_WRITE ||
-                wr->type() == BS_HEAP_BIG_INTENT ||
-                wr->type() == BS_HEAP_DELETE)
-            {
-                return false;
-            }
+            break;
         }
-        if (need_skip && wr->type() == BS_HEAP_SMALL_WRITE &&
-            wr->small().offset < blk_end && wr->small().offset+wr->small().len > blk_start)
-        {
-            // Small write may mutate big write checksums during flush
-            skip_csum = COPY_BUF_SKIP_CSUM;
-        }
-        return true;
-    });
-    if (need_wait)
-    {
-undo_wait:
-        // Need to wait. undo added requests, unlock lsn
-        heap->unlock_entry(op->oid);
-        free_read_buffers(rv);
-        rv.clear();
-        return 0;
     }
     if (!found)
     {
         // May happen if there are entries but all of them are > requested version
-        heap->unlock_entry(op->oid);
+        heap->unlock_entry(op->oid, PRIV(op)->lsn);
         op->version = 0;
         op->retval = -ENOENT;
         FINISH_OP(op);
@@ -104,13 +55,17 @@ undo_wait:
     assert(fulfilled == op->len);
     if (!fulfill_read(op))
     {
-        goto undo_wait;
+        // Need to wait. undo added requests, unlock lsn
+        heap->unlock_entry(op->oid, PRIV(op)->lsn);
+        free_read_buffers(rv);
+        rv.clear();
+        return 0;
     }
     op->version = result_version;
     if (!PRIV(op)->pending_ops)
     {
         // everything is fulfilled from memory
-        heap->unlock_entry(op->oid);
+        heap->unlock_entry(op->oid, PRIV(op)->lsn);
         op->retval = op->len;
         free_read_buffers(rv);
         FINISH_OP(op);
@@ -153,20 +108,20 @@ int blockstore_impl_t::fulfill_read(blockstore_op_t *op)
     return 1;
 }
 
-uint32_t blockstore_impl_t::prepare_read(std::vector<copy_buffer_t> & read_vec, heap_entry_t *obj, heap_entry_t *wr, uint32_t start, uint32_t end, uint32_t skip_csum)
+uint32_t blockstore_impl_t::prepare_read(std::vector<copy_buffer_t> & read_vec, heap_object_t *obj, heap_write_t *wr, uint32_t start, uint32_t end)
 {
-    if (wr->type() == BS_HEAP_BIG_WRITE || wr->type() == BS_HEAP_BIG_INTENT)
+    if (wr->type() == BS_HEAP_BIG_WRITE)
     {
-        return prepare_read_with_bitmaps(read_vec, obj, wr, start, end, skip_csum);
+        return prepare_read_with_bitmaps(read_vec, obj, wr, start, end);
     }
-    if (wr->type() == BS_HEAP_DELETE)
+    if (wr->type() == BS_HEAP_TOMBSTONE)
     {
         return prepare_read_zero(read_vec, start, end);
     }
-    return prepare_read_simple(read_vec, obj, wr, start, end, skip_csum);
+    return prepare_read_simple(read_vec, obj, wr, start, end);
 }
 
-uint32_t blockstore_impl_t::prepare_read_with_bitmaps(std::vector<copy_buffer_t> & read_vec, heap_entry_t *obj, heap_entry_t *wr, uint32_t start, uint32_t end, uint32_t skip_csum)
+uint32_t blockstore_impl_t::prepare_read_with_bitmaps(std::vector<copy_buffer_t> & read_vec, heap_object_t *obj, heap_write_t *wr, uint32_t start, uint32_t end)
 {
     // BIG_WRITEs contain a bitmap and we have to handle its holes
     uint32_t res = 0;
@@ -189,7 +144,7 @@ uint32_t blockstore_impl_t::prepare_read_with_bitmaps(std::vector<copy_buffer_t>
         }
         if (bmp_end > bmp_start)
         {
-            res += prepare_read_simple(read_vec, obj, wr, bmp_start * dsk.bitmap_granularity, bmp_end * dsk.bitmap_granularity, skip_csum);
+            res += prepare_read_simple(read_vec, obj, wr, bmp_start * dsk.bitmap_granularity, bmp_end * dsk.bitmap_granularity);
             bmp_start = bmp_end;
         }
     }
@@ -211,7 +166,7 @@ uint32_t blockstore_impl_t::prepare_read_zero(std::vector<copy_buffer_t> & read_
     return res;
 }
 
-uint32_t blockstore_impl_t::prepare_read_simple(std::vector<copy_buffer_t> & read_vec, heap_entry_t *obj, heap_entry_t *wr, uint32_t start, uint32_t end, uint32_t skip_csum)
+uint32_t blockstore_impl_t::prepare_read_simple(std::vector<copy_buffer_t> & read_vec, heap_object_t *obj, heap_write_t *wr, uint32_t start, uint32_t end)
 {
     uint32_t res = 0;
     if (wr->type() == BS_HEAP_SMALL_WRITE || wr->type() == BS_HEAP_INTENT_WRITE)
@@ -235,7 +190,7 @@ uint32_t blockstore_impl_t::prepare_read_simple(std::vector<copy_buffer_t> & rea
                 .disk_offset = start,
                 .disk_len = end-start,
                 .buf = buffer_area + wr->small().location + start - wr->small().offset,
-                .wr = wr,
+                .wr_lsn = wr->lsn,
             });
         }
         else if (dsk.csum_block_size <= dsk.bitmap_granularity)
@@ -254,6 +209,14 @@ uint32_t blockstore_impl_t::prepare_read_simple(std::vector<copy_buffer_t> & rea
                 blk_start = blk_start < wr->small().offset ? wr->small().offset : blk_start;
                 blk_end = blk_end > wr->small().offset+wr->small().len ? wr->small().offset+wr->small().len : blk_end;
             }
+            uint32_t skip_csum = 0;
+            if (!perfect_csum_update && wr->type() == BS_HEAP_BIG_WRITE)
+            {
+                for (auto owr = obj->get_writes(); owr && owr != wr; owr = owr->next())
+                    if ((owr->type() == BS_HEAP_INTENT_WRITE || owr->type() == BS_HEAP_SMALL_WRITE) &&
+                        owr->small().offset < blk_end && owr->small().offset+owr->small().len > blk_start)
+                        skip_csum = COPY_BUF_SKIP_CSUM;
+            }
             if ((blk_end-1)/dsk.csum_block_size == blk_start/dsk.csum_block_size ||
                 blk_end/dsk.csum_block_size == blk_start/dsk.csum_block_size+1 && blk_end != end && blk_start != start ||
                 blk_end == end && blk_start == start)
@@ -280,37 +243,21 @@ uint32_t blockstore_impl_t::prepare_read_simple(std::vector<copy_buffer_t> & rea
     return res;
 }
 
-void blockstore_impl_t::prepare_disk_read(std::vector<copy_buffer_t> & read_vec, int pos, heap_entry_t *obj, heap_entry_t *wr,
+void blockstore_impl_t::prepare_disk_read(std::vector<copy_buffer_t> & read_vec, int pos, heap_object_t *obj, heap_write_t *wr,
     uint32_t blk_start, uint32_t blk_end, uint32_t start, uint32_t end, uint32_t copy_flags)
 {
-    uint64_t loc = 0;
-    if (wr->type() == BS_HEAP_INTENT_WRITE)
-    {
-        heap_entry_t *big_wr = wr;
-        while (big_wr && big_wr->type() == BS_HEAP_INTENT_WRITE)
-        {
-            big_wr = heap->prev(big_wr);
-        }
-        assert(big_wr->type() == BS_HEAP_BIG_WRITE || big_wr->type() == BS_HEAP_BIG_INTENT);
-        loc = big_wr->big_location(heap);
-    }
-    else if (wr->type() == BS_HEAP_SMALL_WRITE)
-    {
-        loc = wr->small().location-wr->small().offset;
-    }
-    else
-    {
-        assert(wr->type() == BS_HEAP_BIG_WRITE || wr->type() == BS_HEAP_BIG_INTENT);
-        loc = wr->big_location(heap);
-    }
+    // Only one INTENT_WRITE is allowed at a time
+    assert(wr->type() != BS_HEAP_INTENT_WRITE || wr->next()->type() == BS_HEAP_BIG_WRITE);
     copy_buffer_t vec = {
         .copy_flags = (wr->type() == BS_HEAP_SMALL_WRITE ? COPY_BUF_JOURNAL : COPY_BUF_DATA) | copy_flags,
         .offset = start,
         .len = end-start,
-        .disk_loc = loc,
+        .disk_loc = (wr->type() == BS_HEAP_INTENT_WRITE ? wr->next()->big_location(heap)
+            : (wr->type() == BS_HEAP_SMALL_WRITE ? wr->small().location-wr->small().offset
+            : wr->big_location(heap))),
         .disk_offset = blk_start,
         .disk_len = blk_end - blk_start,
-        .wr = wr,
+        .wr_lsn = wr->lsn,
     };
     if (blk_start != start || blk_end != end)
     {
@@ -318,7 +265,7 @@ void blockstore_impl_t::prepare_disk_read(std::vector<copy_buffer_t> & read_vec,
         vec.copy_flags |= COPY_BUF_PADDED;
         if (pos > 0 && read_vec.size() >= pos &&
             read_vec[pos-1].copy_flags == vec.copy_flags &&
-            read_vec[pos-1].wr == vec.wr &&
+            read_vec[pos-1].wr_lsn == vec.wr_lsn &&
             read_vec[pos-1].disk_offset <= vec.disk_offset &&
             read_vec[pos-1].disk_offset+read_vec[pos-1].disk_len >= blk_end)
         {
@@ -402,7 +349,7 @@ void blockstore_impl_t::handle_read_event(ring_data_t *data, blockstore_op_t *op
             op->retval = -EDOM;
         else if (op->retval == 0)
             op->retval = op->len;
-        heap->unlock_entry(op->oid);
+        heap->unlock_entry(op->oid, PRIV(op)->lsn);
         free_read_buffers(PRIV(op)->read_vec);
         FINISH_OP(op);
     }
@@ -410,15 +357,7 @@ void blockstore_impl_t::handle_read_event(ring_data_t *data, blockstore_op_t *op
 
 bool blockstore_impl_t::verify_read_checksums(blockstore_op_t *op)
 {
-    bool skip_all = false;
-    if (PRIV(op)->lsn)
-    {
-        heap_entry_t *obj = heap->read_entry(op->oid);
-        if (obj->lsn != PRIV(op)->lsn) // check top lsn
-        {
-            skip_all = true;
-        }
-    }
+    heap_object_t *obj = heap->read_locked_entry(op->oid, PRIV(op)->lsn);
     auto & rv = PRIV(op)->read_vec;
     for (auto & vec: rv)
     {
@@ -426,22 +365,23 @@ bool blockstore_impl_t::verify_read_checksums(blockstore_op_t *op)
             continue;
         if (vec.copy_flags & COPY_BUF_PADDED)
             memcpy(op->buf + vec.offset - op->offset, vec.buf + vec.offset - vec.disk_offset, vec.len);
-        if (vec.copy_flags & (COPY_BUF_COALESCED|COPY_BUF_SKIP_CSUM) ||
-            skip_all && vec.wr->type() != BS_HEAP_SMALL_WRITE)
+        if (vec.copy_flags & (COPY_BUF_COALESCED|COPY_BUF_SKIP_CSUM))
             continue;
+        heap_write_t *wr = obj->get_writes();
+        while (wr && wr->lsn != vec.wr_lsn)
+            wr = wr->next();
+        assert(wr);
         uint8_t *buf = vec.buf ? vec.buf : (op->buf + vec.offset - op->offset);
-        uint32_t *csums = (uint32_t*)(vec.wr->get_checksums(heap)
+        uint32_t *csums = (uint32_t*)(wr->get_checksums(heap)
             + (vec.disk_offset/dsk.csum_block_size)*(dsk.data_csum_type & 0xFF)
-            - ((vec.wr->type() == BS_HEAP_BIG_WRITE || vec.wr->type() == BS_HEAP_BIG_INTENT)
-                ? 0 : (vec.wr->small().offset/dsk.csum_block_size)*(dsk.data_csum_type & 0xFF)));
-        if (!heap->calc_block_checksums(csums, buf, vec.wr->get_int_bitmap(heap),
+            - ((wr->type() == BS_HEAP_BIG_WRITE) ? 0 : (wr->small().offset/dsk.csum_block_size)*(dsk.data_csum_type & 0xFF)));
+        if (!heap->calc_block_checksums(csums, buf, wr->get_int_bitmap(heap),
             vec.disk_offset, vec.disk_offset+vec.disk_len, false, [&](uint32_t mismatch_pos, uint32_t expected_csum, uint32_t real_csum)
             {
                 printf(
-                    "Checksum mismatch in object %jx:%jx v%ju, offset 0x%x in %s area at offset 0x%jx during read %x+%x: %08x expected vs %08x actual\n",
+                    "Checksum mismatch in object %jx:%jx v%ju, offset 0x%x in %s area at offset 0x%jx: %08x expected vs %08x actual\n",
                     op->oid.inode, op->oid.stripe, op->version, mismatch_pos,
                     (vec.copy_flags & COPY_BUF_JOURNAL) ? "buffer" : "data", vec.disk_loc + vec.disk_offset,
-                    op->offset, op->len,
                     expected_csum, real_csum
                 );
             }))
@@ -454,33 +394,23 @@ bool blockstore_impl_t::verify_read_checksums(blockstore_op_t *op)
 
 int blockstore_impl_t::read_bitmap(object_id oid, uint64_t target_version, void *bitmap, uint64_t *result_version)
 {
-    heap_entry_t *obj = heap->read_entry(oid);
+    heap_object_t *obj = heap->read_entry(oid, NULL);
     if (obj)
     {
-        bool found = false;
-        heap->iterate_with_stable(obj, obj->lsn, [&](heap_entry_t *wr, bool stable)
+        for (auto wr = obj->get_writes(); wr; wr = wr->next())
         {
-            if (target_version >= wr->version)
+            if (target_version < wr->version)
             {
-                if (wr->type() == BS_HEAP_DELETE)
-                {
-                    return false;
-                }
-                found = true;
-                if (result_version)
-                {
-                    *result_version = wr->version;
-                }
-                if (bitmap)
-                {
-                    memcpy(bitmap, wr->get_ext_bitmap(heap), dsk.clean_entry_bitmap_size);
-                }
-                return false;
+                continue;
+            }
+            if (result_version)
+            {
+                *result_version = wr->version;
+            }
+            if (bitmap)
+            {
+                memcpy(bitmap, wr->get_ext_bitmap(heap), dsk.clean_entry_bitmap_size);
             }
-            return true;
-        });
-        if (found)
-        {
             return 0;
         }
     }

src/blockstore/blockstore_stable.cpp

@@ -13,82 +13,73 @@ int blockstore_impl_t::dequeue_stable(blockstore_op_t *op)
     else if (priv->op_state == 2) goto resume_2;
     else if (priv->op_state == 3) goto resume_3;
     else if (priv->op_state == 4) goto resume_4;
-    else if (priv->op_state == 5) goto resume_5;
     assert(!priv->op_state);
+    // Modify in-memory state and assign contiguous LSNs
+    priv->stab_pos = 0;
+    priv->lsn = priv->to_lsn = 0;
     op->retval = 0;
-    PRIV(op)->lsn = 0;
-    priv->modified_block = priv->modified_block2 = UINT32_MAX;
-    for (priv->stab_pos = 0; priv->stab_pos < op->len; priv->stab_pos++)
+    while (priv->stab_pos < op->len)
     {
+        uint32_t modified_block = 0;
+        uint64_t new_lsn = 0;
+        uint64_t new_to_lsn = 0;
+        int res = op->opcode == BS_OP_STABLE
+            ? heap->post_stabilize(v[priv->stab_pos].oid, v[priv->stab_pos].version, &modified_block, &new_lsn, &new_to_lsn)
+            : heap->post_rollback(v[priv->stab_pos].oid, v[priv->stab_pos].version, &new_lsn, &modified_block);
+        if (res != 0)
         {
-            auto obj = heap->read_entry(v[priv->stab_pos].oid);
-            if (!obj)
+            assert(res == ENOENT || res == EBUSY);
+            op->retval = -res;
+        }
+        if (new_lsn)
+        {
+            assert(priv->lsn == 0 || priv->to_lsn == new_lsn-1);
+            if (!priv->lsn)
+                priv->lsn = new_lsn;
+            priv->to_lsn = op->opcode == BS_OP_STABLE ? new_to_lsn : new_lsn;
+        }
+        priv->stab_pos++;
+    }
+    // Submit metadata writes
+    priv->stab_pos = 0;
+resume_1:
+    priv->op_state = 1;
+    while (priv->stab_pos < op->len)
+    {
+        uint32_t block_num = 0;
+        heap_object_t *obj = heap->read_entry(v[priv->stab_pos].oid, &block_num);
+        if (obj)
+        {
+            io_uring_sqe *sqe = get_sqe();
+            if (!sqe)
             {
-                op->retval = -ENOENT;
-                FINISH_OP(op);
-                return 2;
-            }
-            int res = op->opcode == BS_OP_STABLE
-                ? heap->add_commit(obj, v[priv->stab_pos].version, &priv->modified_block2)
-                : heap->add_rollback(obj, v[priv->stab_pos].version, &priv->modified_block2);
-            if (res == EBUSY)
-            {
-                op->retval = -EBUSY;
-                FINISH_OP(op);
-                return 2;
-            }
-            if (res == ENOENT)
-            {
-                op->retval = -ENOENT;
-                FINISH_OP(op);
-                return 2;
-            }
-            if (res == ENOSPC)
-            {
-                if (!heap->get_to_compact_count())
-                {
-                    // no space
-                    op->retval = -ENOSPC;
-                    FINISH_OP(op);
-                    return 2;
-                }
-                if (priv->modified_block2 != UINT32_MAX)
-                {
-                    priv->stab_pos--;
-                    goto resume_1;
-                }
-                priv->wait_for = WAIT_COMPACTION;
-                priv->wait_detail = heap->get_compacted_count();
-                flusher->request_trim();
+                if (priv->pending_ops > 0)
+                    return 1;
+                priv->wait_detail = 1;
+                priv->wait_for = WAIT_SQE;
                 return 0;
             }
-            assert(res == 0);
-        }
-resume_1:
-        if (priv->modified_block != UINT32_MAX && priv->modified_block2 != priv->modified_block)
-        {
-            BS_SUBMIT_CHECK_SQES(1);
-            prepare_meta_block_write(priv->modified_block);
-resume_2:
-            if (meta_block_is_pending(priv->modified_block))
-            {
-                priv->op_state = 2;
-                return 1;
-            }
-        }
-        priv->modified_block = priv->modified_block2;
-        if (priv->stab_pos == op->len-1 && priv->modified_block2 != UINT32_MAX)
-        {
-            priv->modified_block2 = UINT32_MAX;
-            goto resume_1;
+            prepare_meta_block_write(op, block_num, sqe);
         }
+        priv->stab_pos++;
     }
+    if (priv->pending_ops > 0)
+    {
+        priv->op_state = 1;
+        return 1;
+    }
+    // Mark writes as completed to allow compaction
+    for (uint64_t lsn = priv->lsn; lsn <= priv->to_lsn; lsn++)
+    {
+        heap->mark_lsn_completed(lsn);
+    }
+    unsynced_meta_write_count++;
     // Fsync, just because our semantics imply that commit (stabilize) is immediately fsynced
-    priv->op_state = 3;
+    priv->op_state = 2;
+resume_2:
 resume_3:
 resume_4:
-resume_5:
-    int res = do_sync(op, 3);
+    int res = do_sync(op, 2);
     if (res != 2)
     {
         return res;

src/blockstore/blockstore_sync.cpp

@@ -9,7 +9,6 @@ int blockstore_impl_t::continue_sync(blockstore_op_t *op)
     if (!PRIV(op)->op_state)
     {
         op->retval = 0;
-        PRIV(op)->lsn = 0;
     }
     int res = do_sync(op, 0);
     if (res == 2)
@@ -19,19 +18,11 @@ int blockstore_impl_t::continue_sync(blockstore_op_t *op)
     return res;
 }
 
-bool blockstore_impl_t::has_unsynced()
-{
-    bool data = (!dsk.disable_data_fsync && unsynced_data_write_count);
-    bool buffer = (!dsk.disable_journal_fsync && unsynced_buffer_write_count);
-    bool meta = (!dsk.disable_meta_fsync && unsynced_meta_write_count);
-    return data || buffer || meta;
-}
-
 bool blockstore_impl_t::submit_fsyncs(int & wait_count)
 {
-    int n = (unsynced_meta_write_count > 0 && !dsk.disable_meta_fsync) +
-        (unsynced_buffer_write_count > 0 && !dsk.disable_journal_fsync && dsk.journal_fd != dsk.meta_fd) +
-        (unsynced_data_write_count > 0 && !dsk.disable_data_fsync && dsk.data_fd != dsk.meta_fd && dsk.data_fd != dsk.journal_fd);
+    int n = ((unsynced_small_write_count > 0 || unsynced_big_write_count > 0 || unsynced_meta_write_count > 0) && !dsk.disable_meta_fsync) +
+        (unsynced_small_write_count > 0 && !dsk.disable_journal_fsync && dsk.journal_fd != dsk.meta_fd) +
+        (unsynced_big_write_count > 0 && !dsk.disable_data_fsync && dsk.data_fd != dsk.meta_fd && dsk.data_fd != dsk.journal_fd);
     if (ringloop->space_left() < n)
     {
         return false;
@@ -49,7 +40,7 @@ bool blockstore_impl_t::submit_fsyncs(int & wait_count)
         if (!wait_count)
             ringloop->wakeup();
     };
-    if (unsynced_meta_write_count > 0 && !dsk.disable_meta_fsync)
+    if ((unsynced_small_write_count > 0 || unsynced_big_write_count > 0 || unsynced_meta_write_count > 0) && !dsk.disable_meta_fsync)
     {
         // fsync meta
         io_uring_sqe *sqe = get_sqe();
@@ -60,7 +51,7 @@ bool blockstore_impl_t::submit_fsyncs(int & wait_count)
         data->callback = cb;
         wait_count++;
     }
-    if (unsynced_buffer_write_count > 0 && !dsk.disable_journal_fsync && dsk.meta_fd != dsk.journal_fd)
+    if (unsynced_small_write_count > 0 && !dsk.disable_journal_fsync && dsk.meta_fd != dsk.journal_fd)
     {
         // fsync buffer
         io_uring_sqe *sqe = get_sqe();
@@ -71,7 +62,7 @@ bool blockstore_impl_t::submit_fsyncs(int & wait_count)
         data->callback = cb;
         wait_count++;
     }
-    if (unsynced_data_write_count > 0 && !dsk.disable_data_fsync && dsk.data_fd != dsk.meta_fd && dsk.data_fd != dsk.journal_fd)
+    if (unsynced_big_write_count > 0 && !dsk.disable_data_fsync && dsk.data_fd != dsk.meta_fd && dsk.data_fd != dsk.journal_fd)
     {
         // fsync data
         io_uring_sqe *sqe = get_sqe();
@@ -82,8 +73,8 @@ bool blockstore_impl_t::submit_fsyncs(int & wait_count)
         data->callback = cb;
         wait_count++;
     }
-    unsynced_data_write_count = 0;
-    unsynced_buffer_write_count = 0;
+    unsynced_big_write_count = 0;
+    unsynced_small_write_count = 0;
     unsynced_meta_write_count = 0;
     return true;
 }
@@ -99,13 +90,12 @@ int blockstore_impl_t::do_sync(blockstore_op_t *op, int base_state)
         // Wait for flusher-initiated sync
         return 0;
     }
-    if (!has_unsynced())
+    if (dsk.disable_journal_fsync && dsk.disable_meta_fsync && dsk.disable_data_fsync || !unsynced_big_write_count && !unsynced_small_write_count)
     {
         // We can return immediately because sync only syncs previous writes
-        unsynced_data_write_count = unsynced_buffer_write_count = unsynced_meta_write_count = 0;
+        unsynced_big_write_count = unsynced_small_write_count = unsynced_meta_write_count = 0;
         return 2;
     }
-    assert(!PRIV(op)->lsn);
     PRIV(op)->lsn = heap->get_completed_lsn();
     if (!submit_fsyncs(PRIV(op)->pending_ops))
     {

src/blockstore/blockstore_write.cpp

@@ -5,110 +5,29 @@
 #include "blockstore_internal.h"
 #include "allocator.h"
 
-#define _REDIRECT_INTENT 0x101
-
 bool blockstore_impl_t::enqueue_write(blockstore_op_t *op)
 {
     clock_gettime(CLOCK_REALTIME, &PRIV(op)->tv_begin);
     return true;
 }
 
-void blockstore_impl_t::prepare_meta_block_write(uint32_t modified_block)
+void blockstore_impl_t::prepare_meta_block_write(blockstore_op_t *op, uint64_t modified_block, io_uring_sqe *sqe)
 {
-    if (modified_blocks.find(modified_block) != modified_blocks.end())
-        return;
-    io_uring_sqe *sqe = get_sqe();
-    assert(sqe != NULL);
+    if (!sqe)
+    {
+        sqe = get_sqe();
+        assert(sqe != NULL);
+    }
     ring_data_t *data = ((ring_data_t*)sqe->user_data);
-    uint8_t *buf = (uint8_t*)memalign_or_die(MEM_ALIGNMENT, dsk.meta_block_size);
-    data->iov = (struct iovec){ buf, (size_t)dsk.meta_block_size };
-    data->callback = [this, modified_block](ring_data_t *data)
-    {
-        live = true;
-        if (data->res != data->iov.iov_len)
-        {
-            // FIXME: our state becomes corrupted after a write error. maybe do something better than just die
-            disk_error_abort("data write", data->res, data->iov.iov_len);
-        }
-        auto it = modified_blocks.find(modified_block);
-        assert(it != modified_blocks.end());
-        free(it->second.buf);
-        modified_blocks.erase(it);
-        heap->complete_block_write(modified_block);
-        ringloop->wakeup();
-    };
-    assert(((uint64_t)modified_block+2)*dsk.meta_block_size <= dsk.meta_area_size);
+    data->iov = (struct iovec){ heap->get_meta_block(modified_block), (size_t)dsk.meta_block_size };
+    data->callback = [this, op](ring_data_t *data) { handle_write_event(data, op); };
+    PRIV(op)->pending_ops++;
     io_uring_prep_writev(
-        sqe, dsk.meta_fd, &data->iov, 1, dsk.meta_offset + ((uint64_t)modified_block+1)*dsk.meta_block_size
+        sqe, dsk.meta_fd, &data->iov, 1, dsk.meta_offset + (modified_block+1)*dsk.meta_block_size
     );
-    unsynced_meta_write_count++;
-    pending_modified_blocks.push_back(modified_block);
-    modified_blocks[modified_block] = { .sent = false, .buf = buf };
-}
-
-bool blockstore_impl_t::meta_block_is_pending(uint32_t modified_block)
-{
-    auto mb_it = modified_blocks.find(modified_block);
-    return mb_it != modified_blocks.end();
-}
-
-bool blockstore_impl_t::intent_write_allowed(blockstore_op_t *op, heap_entry_t *obj)
-{
-    // Parallel writes to the same object are forbidden so "one intent at a time" is fulfilled automatically
-    // Intent writes are disabled when metadata fsync is enabled
-    if (!dsk.disable_meta_fsync)
-    {
-        return false;
-    }
-    // Intent writes are only for replication
-    if (op->opcode != BS_OP_WRITE_STABLE)
-    {
-        return false;
-    }
-    // Operation size should be less than or equal to atomic write size
-    if (!op->len || op->len > dsk.atomic_write_size)
-    {
-        return false;
-    }
-    // Intent-writes are disabled if "absolutely correct during compaction" checksum validation algorithm is enabled
-    // We could also do RMW here when perfect_csum_update is enabled, but it's unclear if we need it
-    if (perfect_csum_update && dsk.csum_block_size > dsk.bitmap_granularity &&
-        ((op->offset % dsk.csum_block_size) || (op->len % dsk.csum_block_size)))
-    {
-        return false;
-    }
-    bool ok = true;
-    heap->iterate_with_stable(obj, obj->lsn, [&](heap_entry_t *wr, bool stable)
-    {
-        // Intent writes are not allowed over buffered writes
-        auto t = wr->type();
-        if (t == BS_HEAP_SMALL_WRITE)
-        {
-            ok = false;
-            return false;
-        }
-        // Intent writes are not allowed over unstable writes
-        if (!stable)
-        {
-            ok = false;
-            return false;
-        }
-        // Intent writes are not allowed over unfinished intent writes
-        if ((t == BS_HEAP_INTENT_WRITE || t == BS_HEAP_BIG_INTENT) && wr->lsn > heap->get_fsynced_lsn())
-        {
-            ok = false;
-            return false;
-        }
-        // Intent writes are allowed over BIG_WRITEs even with fsyncs because BIG_WRITE is always counted as fsynced
-        if (t == BS_HEAP_BIG_WRITE || t == BS_HEAP_BIG_INTENT)
-        {
-            return false;
-        }
-        return true;
-    });
-    return ok;
 }
 
+// First step of the write algorithm: dequeue operation and submit initial write(s)
 int blockstore_impl_t::dequeue_write(blockstore_op_t *op)
 {
     if (PRIV(op)->op_state)
@@ -119,42 +38,40 @@ int blockstore_impl_t::dequeue_write(blockstore_op_t *op)
     {
         return 0;
     }
-    PRIV(op)->modified_block = UINT32_MAX;
-    PRIV(op)->write_type = 0;
-    heap_entry_t *obj = heap->read_entry(op->oid);
+    PRIV(op)->is_big = false;
+    uint32_t modified_block = UINT32_MAX, moved_from_block = UINT32_MAX;
+    heap_object_t *obj = heap->read_entry(op->oid, &modified_block);
     if (op->opcode == BS_OP_DELETE)
     {
         // Delete
-        if (!obj || obj->type() == BS_HEAP_DELETE)
+        if (!obj)
         {
             // Already deleted
             op->retval = 0;
             FINISH_OP(op);
             return 2;
         }
-        PRIV(op)->write_type = BS_HEAP_DELETE;
         BS_SUBMIT_CHECK_SQES(1);
-        int res = heap->add_delete(obj, &PRIV(op)->modified_block);
-        if (res == ENOSPC)
-            goto enospc;
+        int res = heap->post_delete(modified_block, obj, &PRIV(op)->lsn);
         assert(res == 0);
-        prepare_meta_block_write(PRIV(op)->modified_block);
-        PRIV(op)->pending_ops++;
+        prepare_meta_block_write(op, modified_block);
         PRIV(op)->op_state = 5;
         write_iodepth++;
     }
     // FIXME: Allow to do initial writes as buffered, not redirected
     // FIXME: Allow to do direct writes over holes
-    else if (!obj || obj->type() == BS_HEAP_DELETE || op->offset == 0 && op->len == dsk.data_block_size)
+    else if (!obj || obj->get_writes()->type() == BS_HEAP_TOMBSTONE ||
+        op->offset == 0 && op->len == dsk.data_block_size)
     {
         // Big (redirect) write
-        PRIV(op)->write_type = dsk.disable_data_fsync || op->opcode != BS_OP_WRITE_STABLE ? BS_HEAP_BIG_WRITE : _REDIRECT_INTENT;
         BS_SUBMIT_CHECK_SQES(1);
-        PRIV(op)->location = heap->find_free_data();
-        if (PRIV(op)->location == UINT64_MAX)
+        PRIV(op)->is_big = true;
+        uint32_t tmp_block;
+        uint64_t loc = heap->find_free_data();
+        if (loc == UINT64_MAX ||
+            !obj && heap->get_block_for_new_object(tmp_block) != 0)
         {
-enospc:
-            if (!heap->get_to_compact_count())
+            if (!heap->get_inflight_queue_size())
             {
                 // no space
                 op->retval = -ENOSPC;
@@ -162,11 +79,11 @@ enospc:
                 return 2;
             }
             PRIV(op)->wait_for = WAIT_COMPACTION;
-            PRIV(op)->wait_detail = heap->get_compacted_count();
+            PRIV(op)->wait_detail = flusher->get_compact_counter();
             flusher->request_trim();
             return 0;
         }
-        uint64_t loc = PRIV(op)->location;
+        PRIV(op)->location = loc;
 #ifdef BLOCKSTORE_DEBUG
         printf(
             "Allocate offset %ju for %jx:%jx v%ju\n",
@@ -174,61 +91,103 @@ enospc:
         );
 #endif
         heap->use_data(op->oid.inode, PRIV(op)->location);
+        if (!dsk.disable_data_fsync && dsk.disable_meta_fsync)
+        {
+            // Do big_write as an INTENT to avoid data fsync
+            int res = make_big_write(op, 0, 0, &modified_block, &moved_from_block);
+            assert(res == 0);
+            if (moved_from_block != UINT32_MAX)
+                prepare_meta_block_write(op, moved_from_block);
+            obj = heap->read_entry(op->oid, &modified_block);
+            heap->mark_lsn_completed(PRIV(op)->lsn);
+            goto process_intent;
+        }
         io_uring_sqe *sqe = get_sqe();
         ring_data_t *data = ((ring_data_t*)sqe->user_data);
-        data->iov = (struct iovec){ op->buf, op->len };
-        data->callback = [this, op](ring_data_t *data) { handle_write_event(data, op); };
-        assert(loc+op->offset+op->len <= dsk.block_count*dsk.data_block_size);
-        io_uring_prep_writev(sqe, dsk.data_fd, &data->iov, 1, dsk.data_offset + loc + op->offset);
-        PRIV(op)->pending_ops++;
-        write_iodepth++;
-        if (PRIV(op)->write_type == BS_HEAP_BIG_WRITE)
+        uint64_t stripe_offset = (op->offset % dsk.bitmap_granularity);
+        uint64_t stripe_end = (op->offset + op->len) % dsk.bitmap_granularity;
+        // Zero fill up to dsk.bitmap_granularity
+        int vcnt = 0;
+        if (stripe_offset)
         {
-            PRIV(op)->op_state = 1;
-            inflight_big++;
+            PRIV(op)->iov_zerofill[vcnt++] = (struct iovec){ zero_object, (size_t)stripe_offset };
         }
-        else
-            PRIV(op)->op_state = 3;
+        PRIV(op)->iov_zerofill[vcnt++] = (struct iovec){ op->buf, op->len };
+        if (stripe_end)
+        {
+            stripe_end = dsk.bitmap_granularity - stripe_end;
+            PRIV(op)->iov_zerofill[vcnt++] = (struct iovec){ zero_object, (size_t)stripe_end };
+        }
+        data->iov.iov_len = op->len + stripe_offset + stripe_end; // to check it in the callback
+        data->callback = [this, op](ring_data_t *data) { handle_write_event(data, op); };
+        io_uring_prep_writev(
+            sqe, dsk.data_fd, PRIV(op)->iov_zerofill, vcnt, dsk.data_offset + loc + op->offset - stripe_offset
+        );
+        PRIV(op)->pending_ops++;
+        PRIV(op)->op_state = 1;
+        write_iodepth++;
+        inflight_big++;
     }
-    else if (intent_write_allowed(op, obj))
+    // Only one INTENT_WRITE is allowed at a time, but in fact,
+    // parallel writes to the same object are forbidden anyway
+    else if (op->len > 0 && op->len <= dsk.atomic_write_size &&
+        // Intent-writes are disabled if "absolutely correct during compaction" checksum validation algorithm is enabled
+        // We could also do RMW here when perfect_csum_update is enabled, but it's unclear if we need it
+        (!perfect_csum_update || dsk.csum_block_size <= dsk.bitmap_granularity ||
+            !(op->offset % dsk.csum_block_size) &&
+            !(op->len % dsk.csum_block_size) &&
+            (obj->get_writes()->entry_type != (BS_HEAP_INTENT_WRITE|BS_HEAP_STABLE) ||
+            obj->get_writes()->can_be_collapsed(heap))) &&
+        // One intent-write is allowed even with fsyncs because BIG_WRITE is always counted as fsynced
+        dsk.disable_meta_fsync &&
+        (op->opcode == BS_OP_WRITE_STABLE &&
+            (obj->get_writes()->entry_type == (BS_HEAP_BIG_WRITE|BS_HEAP_STABLE) ||
+            obj->get_writes()->entry_type == (BS_HEAP_INTENT_WRITE|BS_HEAP_STABLE) && dsk.disable_data_fsync) ||
+        op->opcode == BS_OP_WRITE && obj->get_writes()->entry_type == BS_HEAP_BIG_WRITE))
     {
         // Direct intent-write
         BS_SUBMIT_CHECK_SQES(1);
-        int res = 0;
-        if (dsk.csum_block_size <= dsk.bitmap_granularity &&
-            (obj->entry_type == (BS_HEAP_BIG_INTENT|BS_HEAP_STABLE) ||
-            obj->entry_type == (BS_HEAP_BIG_WRITE|BS_HEAP_STABLE)))
+        if (obj->get_writes()->type() == BS_HEAP_BIG_WRITE)
         {
-            // Even more simplified BIG_INTENT writes
-            // FIXME: Support RMW mode for csum_block_size > bitmap_granularity
-            PRIV(op)->write_type = BS_HEAP_BIG_INTENT;
-            PRIV(op)->location = obj->big_location(heap);
-            res = heap->add_big_intent(op->oid, &obj, op->version, op->offset, op->len, op->bitmap,
-                (uint8_t*)op->buf, NULL, &PRIV(op)->modified_block);
-            if (res == ENOSPC)
-                goto enospc;
-            assert(res == 0);
-            PRIV(op)->lsn = obj->lsn;
+            PRIV(op)->location = obj->get_writes()->big_location(heap);
         }
         else
         {
-            PRIV(op)->write_type = BS_HEAP_INTENT_WRITE;
-            auto wr = obj;
-            while (wr && (wr->type() == BS_HEAP_INTENT_WRITE || wr->type() == BS_HEAP_COMMIT || wr->type() == BS_HEAP_ROLLBACK))
-            {
-                wr = heap->prev(wr);
-            }
-            assert(wr && (wr->type() == BS_HEAP_BIG_WRITE || wr->type() == BS_HEAP_BIG_INTENT));
-            PRIV(op)->location = wr->big_location(heap);
-            res = heap->add_small_write(op->oid, &obj, (BS_HEAP_INTENT_WRITE | (op->opcode == BS_OP_WRITE_STABLE ? BS_HEAP_STABLE : 0)),
-                op->version, op->offset, op->len, 0, op->bitmap, (uint8_t*)op->buf, &PRIV(op)->modified_block);
-            if (res == ENOSPC)
-                goto enospc;
-            assert(res == 0);
-            PRIV(op)->lsn = obj->lsn;
+            assert(obj->get_writes()->next()->type() == BS_HEAP_BIG_WRITE);
+            PRIV(op)->location = obj->get_writes()->next()->big_location(heap);
         }
-        prepare_meta_block_write(PRIV(op)->modified_block);
-        PRIV(op)->pending_ops++;
+process_intent:
+        uint8_t wr_buf[heap->get_max_write_entry_size()];
+        heap_write_t *wr = (heap_write_t*)wr_buf;
+        wr->version = op->version;
+        wr->entry_type = BS_HEAP_INTENT_WRITE | (op->opcode == BS_OP_WRITE_STABLE ? BS_HEAP_STABLE : 0);
+        wr->small().offset = op->offset;
+        wr->small().len = op->len;
+        wr->small().location = 0;
+        if (op->bitmap)
+            memcpy(wr->get_ext_bitmap(heap), op->bitmap, dsk.clean_entry_bitmap_size);
+        heap->calc_checksums(wr, (uint8_t*)op->buf, true);
+        int res = heap->post_write(modified_block, op->oid, obj, wr, &moved_from_block);
+        if (res == EAGAIN)
+        {
+            assert(heap->get_inflight_queue_size());
+            PRIV(op)->wait_for = WAIT_COMPACTION;
+            PRIV(op)->wait_detail = flusher->get_compact_counter();
+            flusher->request_trim();
+            return 0;
+        }
+        else if (res == ENOSPC)
+        {
+            // no space
+            op->retval = -ENOSPC;
+            FINISH_OP(op);
+            return 2;
+        }
+        assert(res == 0);
+        PRIV(op)->lsn = wr->lsn;
+        if (moved_from_block != UINT32_MAX)
+            prepare_meta_block_write(op, moved_from_block);
+        prepare_meta_block_write(op, modified_block);
         PRIV(op)->op_state = 9;
         write_iodepth++;
     }
@@ -236,29 +195,53 @@ enospc:
     {
         // Small (buffered) overwrite
         // First check if there is free buffer space
-        PRIV(op)->write_type = BS_HEAP_SMALL_WRITE;
         uint64_t loc = !op->len ? 0 : heap->find_free_buffer_area(op->len);
         if (loc == UINT64_MAX)
         {
             PRIV(op)->wait_for = WAIT_COMPACTION;
-            PRIV(op)->wait_detail = heap->get_compacted_count();
+            PRIV(op)->wait_detail = flusher->get_compact_counter();
             flusher->request_trim();
             return 0;
         }
         // There is sufficient space. Check SQE(s)
         BS_SUBMIT_CHECK_SQES(1 + (op->len > 0 ? 1 : 0));
-        int res = heap->add_small_write(op->oid, &obj, (BS_HEAP_SMALL_WRITE | (op->opcode == BS_OP_WRITE_STABLE ? BS_HEAP_STABLE : 0)),
-            op->version, op->offset, op->len, loc, op->bitmap, (uint8_t*)op->buf, &PRIV(op)->modified_block);
-        if (res == ENOSPC)
-            goto enospc;
+        uint8_t wr_buf[heap->get_max_write_entry_size()];
+        heap_write_t *wr = (heap_write_t*)wr_buf;
+        wr->version = op->version;
+        wr->entry_type = BS_HEAP_SMALL_WRITE | (op->opcode == BS_OP_WRITE_STABLE ? BS_HEAP_STABLE : 0);
+        wr->small().offset = op->offset;
+        wr->small().len = op->len;
+        wr->small().location = loc;
+        PRIV(op)->location = loc;
+        if (op->bitmap)
+            memcpy(wr->get_ext_bitmap(heap), op->bitmap, dsk.clean_entry_bitmap_size);
+        heap->calc_checksums(wr, (uint8_t*)op->buf, true);
+        int res = heap->post_write(modified_block, op->oid, obj, wr, &moved_from_block);
+        if (res == EAGAIN)
+        {
+            assert(heap->get_inflight_queue_size());
+            PRIV(op)->wait_for = WAIT_COMPACTION;
+            PRIV(op)->wait_detail = flusher->get_compact_counter();
+            flusher->request_trim();
+            return 0;
+        }
+        else if (res == ENOSPC)
+        {
+            // no space
+            op->retval = -ENOSPC;
+            FINISH_OP(op);
+            return 2;
+        }
         assert(res == 0);
-        PRIV(op)->lsn = obj->lsn;
-        prepare_meta_block_write(PRIV(op)->modified_block);
-        PRIV(op)->pending_ops++;
+        PRIV(op)->lsn = wr->lsn;
+        if (op->len)
+            heap->use_buffer_area(op->oid.inode, loc, op->len);
+        if (moved_from_block != UINT32_MAX)
+            prepare_meta_block_write(op, moved_from_block);
+        prepare_meta_block_write(op, modified_block);
         if (op->len > 0)
         {
             // Prepare buffered data write
-            heap->use_buffer_area(op->oid.inode, loc, op->len);
             if (dsk.inmemory_journal)
             {
                 memcpy((uint8_t*)buffer_area + loc, op->buf, op->len);
@@ -266,7 +249,6 @@ enospc:
             BS_SUBMIT_GET_SQE(sqe2, data2);
             data2->iov = (struct iovec){ op->buf, op->len };
             data2->callback = [this, op](ring_data_t *data) { handle_write_event(data, op); };
-            assert(loc+op->len <= dsk.journal_len);
             io_uring_prep_writev(sqe2, dsk.journal_fd, &data2->iov, 1, dsk.journal_offset + loc);
             PRIV(op)->pending_ops++;
         }
@@ -281,6 +263,26 @@ enospc:
     return 1;
 }
 
+int blockstore_impl_t::make_big_write(blockstore_op_t *op, uint32_t offset, uint32_t len, uint32_t *modified_block, uint32_t *moved_from_block)
+{
+    uint8_t wr_buf[heap->get_max_write_entry_size()];
+    heap_write_t *wr = (heap_write_t*)wr_buf;
+    wr->entry_type = BS_HEAP_BIG_WRITE | (op->opcode == BS_OP_WRITE_STABLE ? BS_HEAP_STABLE : 0);
+    wr->version = op->version;
+    wr->set_big_location(heap, PRIV(op)->location);
+    if (op->bitmap)
+        memcpy(wr->get_ext_bitmap(heap), op->bitmap, dsk.clean_entry_bitmap_size);
+    memset(wr->get_int_bitmap(heap), 0, dsk.clean_entry_bitmap_size);
+    bitmap_set(wr->get_int_bitmap(heap), offset, len, dsk.bitmap_granularity);
+    heap->calc_checksums(wr, (uint8_t*)op->buf, true, offset, len);
+    int res = heap->post_write(op->oid, wr, modified_block, moved_from_block);
+    if (res != 0)
+        return res;
+    assert(res == 0);
+    PRIV(op)->lsn = wr->lsn;
+    return 0;
+}
+
 int blockstore_impl_t::continue_write(blockstore_op_t *op)
 {
     int op_state = PRIV(op)->op_state;
@@ -303,12 +305,6 @@ again:
     {
         // In progress
         assert(op_state < 10);
-        if (PRIV(op)->modified_block != UINT32_MAX &&
-            !meta_block_is_pending(PRIV(op)->modified_block))
-        {
-            PRIV(op)->pending_ops--;
-            PRIV(op)->modified_block = UINT32_MAX;
-        }
         if (PRIV(op)->pending_ops > 0)
             return 1;
         op_state++;
@@ -317,6 +313,7 @@ again:
 resume_2:
     // We must fsync all big writes to avoid complex write workflows
     // It's OK for all HDDs and for server SSDs, but slightly worse for desktop SSDs
+    // The other way is to add another type of MVCC to blockstore_heap: "forward" MVCC :)
     inflight_big--;
     if (!dsk.disable_data_fsync)
     {
@@ -352,44 +349,33 @@ resume_12:
     }
 resume_4:
     {
-        BS_SUBMIT_CHECK_SQES(1);
-        auto obj = heap->read_entry(op->oid);
-        int res = 0;
-        if (PRIV(op)->write_type == _REDIRECT_INTENT)
+        uint32_t modified_block = UINT32_MAX, moved_from_block = UINT32_MAX;
+        int res = make_big_write(op, op->offset, op->len, &modified_block, &moved_from_block);
+        if (res == EAGAIN)
         {
-            res = heap->add_redirect_intent(op->oid, &obj, op->version, op->offset, op->len,
-                PRIV(op)->location, op->bitmap, (uint8_t*)op->buf, &PRIV(op)->modified_block);
-        }
-        else
-        {
-            res = heap->add_big_write(op->oid, obj, op->opcode == BS_OP_WRITE_STABLE,
-                op->version, op->offset, op->len, PRIV(op)->location, op->bitmap, (uint8_t*)op->buf, &PRIV(op)->modified_block);
-        }
-        if (res == ENOSPC)
-        {
-            if (!heap->get_to_compact_count())
-            {
-                // no space
-                heap->free_data(op->oid.inode, PRIV(op)->location);
-                write_iodepth--;
-                op->retval = -ENOSPC;
-                FINISH_OP(op);
-                return 2;
-            }
+            assert(heap->get_inflight_queue_size());
             PRIV(op)->wait_for = WAIT_COMPACTION;
-            PRIV(op)->wait_detail = heap->get_compacted_count();
+            PRIV(op)->wait_detail = flusher->get_compact_counter();
             flusher->request_trim();
-            return 0;
+            return 1;
         }
-        assert(res == 0);
-        prepare_meta_block_write(PRIV(op)->modified_block);
-        PRIV(op)->pending_ops++;
+        else if (res == ENOSPC)
+        {
+            heap->free_data(op->oid.inode, PRIV(op)->location);
+            write_iodepth--;
+            op->retval = -ENOSPC;
+            FINISH_OP(op);
+            return 2;
+        }
+        if (moved_from_block != UINT32_MAX)
+            prepare_meta_block_write(op, moved_from_block);
+        prepare_meta_block_write(op, modified_block);
         PRIV(op)->op_state = 5;
         return 1;
     }
 resume_6:
-    // Apply throttling to not fill the journal too quickly for the SSD+HDD case
-    if (PRIV(op)->write_type == BS_HEAP_SMALL_WRITE && throttle_small_writes)
+    // Apply throttling to not fill the journal too fast for the SSD+HDD case
+    if (!PRIV(op)->is_big && throttle_small_writes)
     {
         // Apply throttling
         timespec tv_end;
@@ -409,12 +395,11 @@ resume_6:
         if (ref_us > exec_us + throttle_threshold_us)
         {
             // Pause reply
-            PRIV(op)->pending_ops++;
             PRIV(op)->op_state = 7;
             // Remember that the timer can in theory be called right here
             tfd->set_timer_us(ref_us-exec_us, false, [this, op](int timer_id)
             {
-                PRIV(op)->pending_ops--;
+                PRIV(op)->op_state = 8;
                 ringloop->wakeup();
             });
             return 1;
@@ -426,27 +411,11 @@ resume_8:
     printf("Ack write %jx:%jx v%ju\n", op->oid.inode, op->oid.stripe, op->version);
 #endif
     op->retval = op->len;
-    if (PRIV(op)->write_type == BS_HEAP_BIG_WRITE)
-    {
-        unsynced_data_write_count++;
-    }
-    else if (PRIV(op)->write_type == BS_HEAP_SMALL_WRITE)
-    {
-        unsynced_buffer_write_count++;
-        heap->complete_lsn_write(PRIV(op)->lsn);
-    }
-    else if (PRIV(op)->write_type == BS_HEAP_BIG_INTENT ||
-        PRIV(op)->write_type == BS_HEAP_INTENT_WRITE)
-    {
-        unsynced_data_write_count++;
-        intent_write_counter++;
-        heap->complete_lsn_write(PRIV(op)->lsn);
-    }
-    else if (PRIV(op)->write_type == _REDIRECT_INTENT)
-    {
-        unsynced_data_write_count++;
-        intent_write_counter++;
-    }
+    heap->mark_lsn_completed(PRIV(op)->lsn);
+    if (PRIV(op)->is_big)
+        unsynced_big_write_count++;
+    else
+        unsynced_small_write_count++;
     write_iodepth--;
     FINISH_OP(op);
     return 2;
@@ -456,10 +425,7 @@ resume_10:
     BS_SUBMIT_GET_SQE(sqe, data);
     data->iov = (struct iovec){ op->buf, op->len };
     data->callback = [this, op](ring_data_t *data) { handle_write_event(data, op); };
-    assert(PRIV(op)->location + op->offset <= dsk.block_count*dsk.data_block_size);
     io_uring_prep_writev(sqe, dsk.data_fd, &data->iov, 1, dsk.data_offset + PRIV(op)->location + op->offset);
-    if (dsk.use_atomic_flag)
-        sqe->rw_flags = RWF_ATOMIC;
     PRIV(op)->pending_ops++;
     PRIV(op)->op_state = 7;
     return 1;

src/blockstore/fio_engine.cpp

@@ -28,7 +28,6 @@
 #include "epoll_manager.h"
 #include "malloc_or_die.h"
 #include "json11/json11.hpp"
-#include "../util/robin_hood.h"
 #include "fio_headers.h"
 
 struct bs_data
@@ -38,15 +37,11 @@ struct bs_data
     ring_loop_t *ringloop;
     /* The list of completed io_u structs. */
     std::vector<io_u*> completed;
-    robin_hood::unordered_flat_map<uint64_t, int> inflight_oids;
-    std::vector<io_u*> postponed;
     int op_n = 0, inflight = 0;
     bool ec = false;
     bool imm = true;
     bool last_sync = false;
     bool trace = false;
-    uint8_t *bitmap = NULL;
-    uint32_t block_size = 0;
 };
 
 struct bs_options
@@ -142,7 +137,6 @@ static void bs_cleanup(struct thread_data *td)
         delete bsd->bs;
         delete bsd->epmgr;
         delete bsd->ringloop;
-        free(bsd->bitmap);
         delete bsd;
     }
 }
@@ -166,12 +160,9 @@ static int bs_init(struct thread_data *td)
                 config[p.first] = p.second.dump();
         }
     }
-    bsd->bitmap = (uint8_t*)malloc_or_die(MAX_DATA_BLOCK_SIZE/512/8);
-    memset(bsd->bitmap, 0, MAX_DATA_BLOCK_SIZE/512/8);
     bsd->ringloop = new ring_loop_t(RINGLOOP_DEFAULT_SIZE);
     bsd->epmgr = new epoll_manager_t(bsd->ringloop);
     bsd->bs = blockstore_i::create(config, bsd->ringloop, bsd->epmgr->tfd);
-    bsd->block_size = bsd->bs->get_block_size();
     bsd->imm = config.find("immediate_commit") == config.end() ||
         config["immediate_commit"] == "all";
     while (1)
@@ -186,33 +177,8 @@ static int bs_init(struct thread_data *td)
     return 0;
 }
 
-static enum fio_q_status _bs_queue(struct thread_data *td, struct io_u *io, bool force);
-
-static void _bs_retry(struct bs_data *bsd, uint64_t offset)
-{
-    // Retry postponed ops
-    auto inflight_it = bsd->inflight_oids.find(offset / bsd->block_size);
-    assert(inflight_it != bsd->inflight_oids.end());
-    inflight_it->second--;
-    if (inflight_it->second > 0)
-    {
-        for (size_t i = 0; i < bsd->postponed.size(); i++)
-        {
-            auto oio = bsd->postponed[i];
-            if (oio->offset/bsd->block_size == offset/bsd->block_size)
-            {
-                bsd->postponed.erase(bsd->postponed.begin()+i);
-                _bs_queue((thread_data*)oio->engine_data, oio, true);
-                break;
-            }
-        }
-    }
-    else
-        bsd->inflight_oids.erase(inflight_it);
-}
-
 /* Begin read or write request. */
-static enum fio_q_status _bs_queue(struct thread_data *td, struct io_u *io, bool force)
+static enum fio_q_status bs_queue(struct thread_data *td, struct io_u *io)
 {
     bs_data *bsd = (bs_data*)td->io_ops_data;
     if (io->ddir == DDIR_SYNC && bsd->last_sync)
@@ -222,22 +188,10 @@ static enum fio_q_status _bs_queue(struct thread_data *td, struct io_u *io, bool
 
     fio_ro_check(td, io);
 
-    io->engine_data = td;
+    io->engine_data = bsd;
 
     if (io->ddir == DDIR_WRITE || io->ddir == DDIR_READ)
-        assert(io->xfer_buflen <= bsd->block_size);
-
-    uint64_t stripe = io->offset / bsd->block_size;
-    if (!force && io->ddir == DDIR_WRITE)
-    {
-        auto & inflight = bsd->inflight_oids[stripe];
-        inflight++;
-        if (inflight > 1)
-        {
-            bsd->postponed.push_back(io);
-            return FIO_Q_QUEUED;
-        }
-    }
+        assert(io->xfer_buflen <= bsd->bs->get_block_size());
 
     blockstore_op_t *op = new blockstore_op_t;
     op->callback = NULL;
@@ -249,16 +203,15 @@ static enum fio_q_status _bs_queue(struct thread_data *td, struct io_u *io, bool
         op->buf = (uint8_t*)io->xfer_buf;
         op->oid = {
             .inode = 1,
-            .stripe = stripe,
+            .stripe = io->offset / bsd->bs->get_block_size(),
         };
         op->version = UINT64_MAX; // last unstable
-        op->offset = io->offset % bsd->block_size;
+        op->offset = io->offset % bsd->bs->get_block_size();
         op->len = io->xfer_buflen;
-        op->bitmap = bsd->bitmap;
         op->callback = [io, n = bsd->op_n](blockstore_op_t *op)
         {
             io->error = op->retval < 0 ? -op->retval : 0;
-            bs_data *bsd = ((bs_data*)((thread_data*)io->engine_data)->io_ops_data);
+            bs_data *bsd = (bs_data*)io->engine_data;
             bsd->inflight--;
             bsd->completed.push_back(io);
             if (bsd->trace)
@@ -271,17 +224,16 @@ static enum fio_q_status _bs_queue(struct thread_data *td, struct io_u *io, bool
         op->buf = (uint8_t*)io->xfer_buf;
         op->oid = {
             .inode = 1,
-            .stripe = stripe,
+            .stripe = io->offset / bsd->bs->get_block_size(),
         };
         op->version = 0; // assign automatically
-        op->offset = io->offset % bsd->block_size;
+        op->offset = io->offset % bsd->bs->get_block_size();
         op->len = io->xfer_buflen;
-        op->bitmap = bsd->bitmap;
         if (bsd->ec)
         {
             op->callback = [io, n = bsd->op_n](blockstore_op_t *op)
             {
-                bs_data *bsd = ((bs_data*)((thread_data*)io->engine_data)->io_ops_data);
+                bs_data *bsd = (bs_data*)io->engine_data;
                 if (bsd->trace)
                     printf("--- OP_WRITE %zx n=%d retval=%d\n", (size_t)op, n, op->retval);
                 if (op->retval < 0)
@@ -289,7 +241,6 @@ static enum fio_q_status _bs_queue(struct thread_data *td, struct io_u *io, bool
                     io->error = op->retval < 0 ? -op->retval : 0;
                     bsd->inflight--;
                     bsd->completed.push_back(io);
-                    _bs_retry(bsd, io->offset);
                     delete op;
                 }
                 else
@@ -303,13 +254,12 @@ static enum fio_q_status _bs_queue(struct thread_data *td, struct io_u *io, bool
                     stab_op->len = 1;
                     stab_op->callback = [io, n](blockstore_op_t *op)
                     {
-                        bs_data *bsd = ((bs_data*)((thread_data*)io->engine_data)->io_ops_data);
+                        bs_data *bsd = (bs_data*)io->engine_data;
                         if (bsd->trace)
                             printf("--- OP_STABLE %zx n=%d retval=%d\n", (size_t)op, n, op->retval);
                         io->error = op->retval < 0 ? -op->retval : 0;
                         bsd->inflight--;
                         bsd->completed.push_back(io);
-                        _bs_retry(bsd, io->offset);
                         delete op;
                     };
                     bsd->bs->enqueue_op(stab_op);
@@ -321,13 +271,12 @@ static enum fio_q_status _bs_queue(struct thread_data *td, struct io_u *io, bool
         {
             op->callback = [io, n = bsd->op_n](blockstore_op_t *op)
             {
-                bs_data *bsd = ((bs_data*)((thread_data*)io->engine_data)->io_ops_data);
+                bs_data *bsd = (bs_data*)io->engine_data;
                 if (bsd->trace)
                     printf("--- OP_WRITE_STABLE %zx n=%d retval=%d\n", (size_t)op, n, op->retval);
                 io->error = op->retval < 0 ? -op->retval : 0;
                 bsd->inflight--;
                 bsd->completed.push_back(io);
-                _bs_retry(bsd, io->offset);
                 delete op;
             };
         }
@@ -337,7 +286,7 @@ static enum fio_q_status _bs_queue(struct thread_data *td, struct io_u *io, bool
         op->opcode = BS_OP_SYNC;
         op->callback = [io, n = bsd->op_n](blockstore_op_t *op)
         {
-            bs_data *bsd = ((bs_data*)((thread_data*)io->engine_data)->io_ops_data);
+            bs_data *bsd = (bs_data*)io->engine_data;
             io->error = op->retval < 0 ? -op->retval : 0;
             bsd->completed.push_back(io);
             bsd->inflight--;
@@ -365,11 +314,6 @@ static enum fio_q_status _bs_queue(struct thread_data *td, struct io_u *io, bool
     return FIO_Q_QUEUED;
 }
 
-static enum fio_q_status bs_queue(struct thread_data *td, struct io_u *io)
-{
-    return _bs_queue(td, io, false);
-}
-
 static int bs_getevents(struct thread_data *td, unsigned int min, unsigned int max, const struct timespec *t)
 {
     bs_data *bsd = (bs_data*)td->io_ops_data;

src/blockstore/multilist.cpp

@@ -300,11 +300,6 @@ uint32_t multilist_index_t::find(uint32_t wanted_used)
     return heads[wanted_used];
 }
 
-uint32_t multilist_index_t::next(uint32_t pos)
-{
-    return nexts[pos];
-}
-
 void multilist_index_t::change(uint32_t pos, uint32_t old_used, uint32_t new_used)
 {
     if (new_used == old_used)

src/blockstore/multilist.h

@@ -32,7 +32,6 @@ struct multilist_index_t
     // used should be always < max_used
     multilist_index_t(uint32_t count, uint32_t max_used, uint32_t init_used);
     uint32_t find(uint32_t wanted_used);
-    uint32_t next(uint32_t pos);
     void change(uint32_t pos, uint32_t old_used, uint32_t new_used);
     void print();
 };

src/blockstore/ondisk_formats.h

@@ -141,7 +141,7 @@ struct __attribute__((__packed__)) journal_entry
 inline uint32_t je_crc32(journal_entry *je)
 {
     // 0x48674bc7 = crc32(4 zero bytes)
-    return je->size < 4 ? 0 : crc32c(0x48674bc7, ((uint8_t*)je)+4, je->size-4);
+    return crc32c(0x48674bc7, ((uint8_t*)je)+4, je->size-4);
 }
 
 // "VITAstor"
@@ -185,8 +185,7 @@ struct __attribute__((__packed__)) blockstore_meta_header_v3_t
     uint32_t data_csum_type;
     uint32_t csum_block_size;
     uint32_t header_csum;
-    uint64_t meta_area_size;
-    uint64_t completed_lsn;
+    uint64_t compacted_lsn;
 
     void set_crc32c();
 };

src/blockstore/v1/flush.cpp

@@ -520,7 +520,6 @@ resume_2:
                 await_sqe(15);
                 data->iov = (struct iovec){ it->buf, (size_t)it->len };
                 data->callback = simple_callback_w;
-                assert(clean_loc+it->offset+it->len <= bs->dsk.block_count*bs->dsk.data_block_size);
                 io_uring_prep_writev(
                     sqe, bs->dsk.data_fd, &data->iov, 1, bs->dsk.data_offset + clean_loc + it->offset
                 );
@@ -750,7 +749,6 @@ bool journal_flusher_co::write_meta_block(flusher_meta_write_t & meta_block, int
     await_sqe(0);
     data->iov = (struct iovec){ meta_block.buf, (size_t)bs->dsk.meta_block_size };
     data->callback = simple_callback_w;
-    assert(bs->dsk.meta_block_size + meta_block.sector + bs->dsk.meta_block_size <= bs->dsk.meta_area_size);
     io_uring_prep_writev(
         sqe, bs->dsk.meta_fd, &data->iov, 1, bs->dsk.meta_offset + bs->dsk.meta_block_size + meta_block.sector
     );

src/blockstore/v1/impl.cpp

@@ -407,77 +407,32 @@ blockstore_clean_db_t& blockstore_impl_t::clean_db_shard(object_id oid)
     return clean_db_shards[(pool_id << (64-POOL_ID_BITS)) | pg_num];
 }
 
-struct bs_reshard_state_t
+void blockstore_impl_t::reshard_clean_db(pool_id_t pool, uint32_t pg_count, uint32_t pg_stripe_size)
 {
-    int state = 0;
-    uint64_t pool_id = 0;
-    uint32_t pg_count = 0;
-    uint32_t pg_stripe_size = 0;
-    uint64_t chunk_size = 0;
-    std::map<pool_pg_id_t, blockstore_clean_db_t> old_shards;
+    uint64_t pool_id = (uint64_t)pool;
     std::map<pool_pg_id_t, blockstore_clean_db_t> new_shards;
-    std::map<pool_pg_id_t, blockstore_clean_db_t>::iterator sh_it;
-    blockstore_clean_db_t::iterator obj_it;
-};
-
-void* blockstore_impl_t::reshard_start(pool_id_t pool, uint32_t pg_count, uint32_t pg_stripe_size, uint64_t chunk_limit)
-{
-    auto & settings = clean_db_settings[pool];
-    if (settings.pg_count == pg_count && settings.pg_stripe_size == pg_stripe_size)
-    {
-        return NULL;
-    }
-    bs_reshard_state_t *st = new bs_reshard_state_t;
-    st->state = 0;
-    st->pool_id = pool;
-    st->pg_count = pg_count;
-    st->pg_stripe_size = pg_stripe_size;
-    auto sh_it = clean_db_shards.lower_bound((st->pool_id << (64-POOL_ID_BITS)));
+    auto sh_it = clean_db_shards.lower_bound((pool_id << (64-POOL_ID_BITS)));
     while (sh_it != clean_db_shards.end() &&
-        (sh_it->first >> (64-POOL_ID_BITS)) == st->pool_id)
+        (sh_it->first >> (64-POOL_ID_BITS)) == pool_id)
     {
-        st->old_shards[sh_it->first] = std::move(sh_it->second);
+        for (auto & pair: sh_it->second)
+        {
+            // like map_to_pg()
+            uint64_t pg_num = (pair.first.stripe / pg_stripe_size) % pg_count + 1;
+            uint64_t shard_id = (pool_id << (64-POOL_ID_BITS)) | pg_num;
+            new_shards[shard_id][pair.first] = pair.second;
+        }
         clean_db_shards.erase(sh_it++);
     }
-    bool finished = reshard_continue(st, chunk_limit);
-    return finished ? NULL : st;
-}
-
-bool blockstore_impl_t::reshard_continue(void *reshard_state, uint64_t chunk_limit)
-{
-    bs_reshard_state_t *st = (bs_reshard_state_t*)reshard_state;
-    uint64_t chunk_size = 0;
-    if (st->state == 1)
-        goto resume_1;
-    for (st->sh_it = st->old_shards.begin(); st->sh_it != st->old_shards.end(); )
-    {
-        for (st->obj_it = st->sh_it->second.begin(); st->obj_it != st->sh_it->second.end(); st->obj_it++)
-        {
-            if (chunk_limit > 0 && chunk_size >= chunk_limit)
-            {
-                st->state = 1;
-                return false;
-            }
-resume_1:
-            // like map_to_pg()
-            uint64_t pg_num = (st->obj_it->first.stripe / st->pg_stripe_size) % st->pg_count + 1;
-            uint64_t shard_id = (st->pool_id << (64-POOL_ID_BITS)) | pg_num;
-            st->new_shards[shard_id][st->obj_it->first] = st->obj_it->second;
-            chunk_size++;
-        }
-        st->old_shards.erase(st->sh_it++);
-    }
-    for (auto sh_it = st->new_shards.begin(); sh_it != st->new_shards.end(); sh_it++)
+    for (sh_it = new_shards.begin(); sh_it != new_shards.end(); sh_it++)
     {
         auto & to = clean_db_shards[sh_it->first];
         to.swap(sh_it->second);
     }
-    clean_db_settings[st->pool_id] = (pool_shard_settings_t){
-        .pg_count = st->pg_count,
-        .pg_stripe_size = st->pg_stripe_size,
+    clean_db_settings[pool_id] = (pool_shard_settings_t){
+        .pg_count = pg_count,
+        .pg_stripe_size = pg_stripe_size,
     };
-    delete st;
-    return true;
 }
 
 void blockstore_impl_t::process_list(blockstore_op_t *op)
@@ -510,10 +465,7 @@ void blockstore_impl_t::process_list(blockstore_op_t *op)
                 sh_it->second.pg_count != pg_count ||
                 sh_it->second.pg_stripe_size != pg_stripe_size)
             {
-                // Sharding mismatch
-                op->retval = -EAGAIN;
-                FINISH_OP(op);
-                return;
+                reshard_clean_db(pool_id, pg_count, pg_stripe_size);
             }
             first_shard = last_shard = ((uint64_t)pool_id << (64-POOL_ID_BITS)) | list_pg;
         }
@@ -855,29 +807,4 @@ std::string blockstore_impl_t::get_op_diag(blockstore_op_t *op)
     return std::string(buf);
 }
 
-uint64_t blockstore_impl_t::get_live_entries()
-{
-    return used_blocks;
-}
-
-uint64_t blockstore_impl_t::get_live_memory()
-{
-    uint64_t used = 0;
-    for (auto & kv: clean_db_shards)
-    {
-        used += kv.second.size() * sizeof(blockstore_clean_db_t::value_type);
-    }
-    return used;
-}
-
-uint64_t blockstore_impl_t::get_garbage_entries()
-{
-    return dirty_db.size();
-}
-
-uint64_t blockstore_impl_t::get_garbage_memory()
-{
-    return (sizeof(obj_ver_id) + sizeof(dirty_entry) + 32) * dirty_db.size();
-}
-
 } // namespace v1

src/blockstore/v1/impl.h

@@ -202,6 +202,7 @@ class blockstore_impl_t: public blockstore_i
     uint8_t* get_clean_entry_bitmap(uint64_t block_loc, int offset);
 
     blockstore_clean_db_t& clean_db_shard(object_id oid);
+    void reshard_clean_db(pool_id_t pool_id, uint32_t pg_count, uint32_t pg_stripe_size);
     void recalc_inode_space_stats(uint64_t pool_id, bool per_inode);
 
     // Journaling
@@ -287,10 +288,6 @@ public:
     void parse_config(blockstore_config_t & config);
     void parse_config(blockstore_config_t & config, bool init);
 
-    // Reshard database for a pool
-    void* reshard_start(pool_id_t pool, uint32_t pg_count, uint32_t pg_stripe_size, uint64_t chunk_limit);
-    bool reshard_continue(void *reshard_state, uint64_t chunk_limit);
-
     // Event loop
     void loop();
 
@@ -332,10 +329,6 @@ public:
     inline uint64_t get_free_block_count() { return dsk.block_count - used_blocks; }
     inline uint32_t get_bitmap_granularity() { return dsk.disk_alignment; }
     inline uint64_t get_journal_size() { return dsk.journal_len; }
-    uint64_t get_live_entries();
-    uint64_t get_live_memory();
-    uint64_t get_garbage_entries();
-    uint64_t get_garbage_memory();
 };
 
 } // namespace v1

src/blockstore/v1/init.cpp

@@ -189,7 +189,7 @@ resume_1:
             printf(
                 "Configuration stored in metadata superblock"
                 " (meta_block_size=%u, data_block_size=%u, bitmap_granularity=%u, data_csum_type=%u, csum_block_size=%u)"
-                " differs from OSD configuration (%ju/%ju/%u, %u/%u).\n",
+                " differs from OSD configuration (%u/%u/%u, %u/%u).\n",
                 hdr->meta_block_size, hdr->data_block_size, hdr->bitmap_granularity,
                 hdr->data_csum_type, hdr->csum_block_size,
                 bs->dsk.meta_block_size, bs->dsk.data_block_size, bs->dsk.bitmap_granularity,