Release 1.10.0

New features: - Implement basic VitastorFS support in [CSI](https://vitastor.io/docs/installation/kubernetes.html) - Implement [NFS RDMA](https://vitastor.io/docs/usage/nfs.html#rdma) support - Pause pool rebalance when monitor detects that it can lead to any OSD becoming full ([osd_backfillfull_ratio](https://vitastor.io/docs/config/monitor.html#osd_backfillfull_ratio)) - Auto-select correct [RDMA device and GID](https://vitastor.io/docs/config/network.html#rdma_device) based on osd_network and RoCEv2 priority - Report slow ops in OSD stats in etcd and show them in vitastor-cli status Bug fixes: - Fix possibly incorrect linked list deserialization in NFS - Fix possible crash in vitastor-nfs --block READDIR operation - Map netlink after forking to show correct PID in vitastor-nbd ls - Simplify and fix create-pool OSD count checks for the case of hosts split into sub-nodes - Make monitor print "Waiting to become master" just once, not every 5s - Take out_size from oimg if not specified in vitastor-cli dd - Do not report OSDs with empty statistics as "full" in status - Trigger double autosync when switching PG state to prevent leaving garbage in non-immediate_commit clusters - Fix a lack of connection timeout for etcd websockets in OSD leading to slower etcd failover (~70s instead of ~10s) - Fix a rare OSD crash during client disconnect - Fix PGs sometimes sticking until OSD restart in the "has_unclean" state with EC pools - Fix metadata partition zeroing in vitastor-disk prepare - Add patches for qemu 9.1 and pve-qemu 9.0 and 9.1 - Fix libvirt 8 patch
Fix clang warnings/errors
2024-12-19 15:49:19 +03:00 · 2024-12-19 15:30:31 +03:00 · 2024-12-19 14:05:12 +03:00 · 2024-12-19 14:05:12 +03:00 · 2024-12-19 14:05:12 +03:00 · 2024-12-19 13:06:47 +03:00
163 changed files with 6465 additions and 1230 deletions
@@ -22,7 +22,7 @@ RUN apt-get update
 RUN apt-get -y install etcd qemu-system-x86 qemu-block-extra qemu-utils fio libasan5 \
    liburing1 liburing-dev 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 -y install jq lp-solve sudo nfs-common
+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; \
@@ -288,6 +288,24 @@ jobs:
          echo ""
        done

+  test_create_halfhost:
+    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_create_halfhost.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_failure_domain:
    runs-on: ubuntu-latest
    needs: build
@@ -828,6 +846,60 @@ jobs:
          echo ""
        done

+  test_resize:
+    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_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_resize_auto:
+    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_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_snapshot_pool2:
+    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_snapshot_pool2.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
@@ -2,6 +2,6 @@ cmake_minimum_required(VERSION 2.8.12)

 project(vitastor)

-set(VITASTOR_VERSION "1.8.0")
+set(VITASTOR_VERSION "1.10.0")

 add_subdirectory(src)
@@ -1,4 +1,4 @@
-## Vitastor
+# Vitastor

 [Read English version](README.md)

@@ -22,7 +22,7 @@ TCP и RDMA и на хорошем железе может достигать з
 Vitastor поддерживает QEMU-драйвер, протоколы NBD и NFS, драйверы OpenStack, OpenNebula, Proxmox, Kubernetes.
 Другие драйверы могут также быть легко реализованы.

-Подробности смотрите в документации по ссылкам ниже.
+Подробности смотрите в документации по ссылкам. Можете начать отсюда: [Быстрый старт](docs/intro/quickstart.ru.md).

 ## Презентации и записи докладов

@@ -51,7 +51,7 @@ Vitastor поддерживает QEMU-драйвер, протоколы NBD и
  - Параметры
    - [Общие](docs/config/common.ru.md)
    - [Сетевые](docs/config/network.ru.md)
-    - [Клиентский код](docs/config/client.en.md)
+    - [Клиентский код](docs/config/client.ru.md)
    - [Глобальные дисковые параметры](docs/config/layout-cluster.ru.md)
    - [Дисковые параметры OSD](docs/config/layout-osd.ru.md)
    - [Прочие параметры OSD](docs/config/osd.ru.md)
@@ -22,7 +22,7 @@ or internal systems of public clouds.
 Vitastor supports QEMU, NBD, NFS protocols, OpenStack, OpenNebula, Proxmox, Kubernetes drivers.
 More drivers may be created easily.

-Read more details below in the documentation.
+Read more details in the documentation. You can start from here: [Quick Start](docs/intro/quickstart.en.md).

 ## Talks and presentations

@@ -22,6 +22,8 @@ RUN apt-get update && \
    (echo "APT::Install-Recommends false;" > /etc/apt/apt.conf) && \
    apt-get update && \
    apt-get install -y e2fsprogs xfsprogs kmod iproute2 \
+        # NFS mount dependencies
+        nfs-common netbase \
        # dependencies of qemu-storage-daemon
        libnuma1 liburing2 libglib2.0-0 libfuse3-3 libaio1 libzstd1 libnettle8 \
        libgmp10 libhogweed6 libp11-kit0 libidn2-0 libunistring2 libtasn1-6 libpcre2-8-0 libffi8 && \
@@ -1,4 +1,4 @@
-VITASTOR_VERSION ?= v1.8.0
+VITASTOR_VERSION ?= v1.10.0

 all: build push

@@ -49,7 +49,7 @@ spec:
            capabilities:
              add: ["SYS_ADMIN"]
            allowPrivilegeEscalation: true
-          image: vitalif/vitastor-csi:v1.8.0
+          image: vitalif/vitastor-csi:v1.10.0
          args:
            - "--node=$(NODE_ID)"
            - "--endpoint=$(CSI_ENDPOINT)"
@@ -121,7 +121,7 @@ spec:
            privileged: true
            capabilities:
              add: ["SYS_ADMIN"]
-          image: vitalif/vitastor-csi:v1.8.0
+          image: vitalif/vitastor-csi:v1.10.0
          args:
            - "--node=$(NODE_ID)"
            - "--endpoint=$(CSI_ENDPOINT)"
@@ -9,8 +9,16 @@ metadata:
 provisioner: csi.vitastor.io
 volumeBindingMode: Immediate
 parameters:
-  etcdVolumePrefix: ""
-  poolId: "1"
+  # CSI driver can create block-based volumes and VitastorFS-based volumes
+  # only VitastorFS-based volumes and raw block volumes (without FS) support ReadWriteMany mode
+  # set this parameter to VitastorFS metadata volume name to use VitastorFS
+  # if unset, block-based volumes will be created
+  vitastorfs: ""
+  # for block-based storage classes, pool ID may be either a string (name) or a number (ID)
+  # for vitastorFS-based storage classes it must be a string - name of the default pool for FS data
+  poolId: "testpool"
+  # volume name prefix for block-based storage classes or NFS subdirectory (including /) for FS-based volumes
+  volumePrefix: ""
  # you can choose other configuration file if you have it in the config map
  # different etcd URLs and prefixes should also be put in the config
  #configPath: "/etc/vitastor/vitastor.conf"
@@ -0,0 +1,25 @@
+---
+apiVersion: storage.k8s.io/v1
+kind: StorageClass
+metadata:
+  namespace: vitastor-system
+  name: vitastor
+  annotations:
+    storageclass.kubernetes.io/is-default-class: "true"
+provisioner: csi.vitastor.io
+volumeBindingMode: Immediate
+parameters:
+  # CSI driver can create block-based volumes and VitastorFS-based volumes
+  # only VitastorFS-based volumes and raw block volumes (without FS) support ReadWriteMany mode
+  # set this parameter to VitastorFS metadata volume name to use VitastorFS
+  # if unset, block-based volumes will be created
+  vitastorfs: "testfs"
+  # for block-based storage classes, pool ID may be either a string (name) or a number (ID)
+  # for vitastorFS-based storage classes it must be a string - name of the default pool for FS data
+  poolId: "testpool"
+  # volume name prefix for block-based storage classes or NFS subdirectory (including /) for FS-based volumes
+  volumePrefix: "k8s/"
+  # you can choose other configuration file if you have it in the config map
+  # different etcd URLs and prefixes should also be put in the config
+  #configPath: "/etc/vitastor/vitastor.conf"
+allowVolumeExpansion: true
@@ -3,10 +3,10 @@ module vitastor.io/csi
 go 1.15

 require (
-	github.com/container-storage-interface/spec v1.4.0
+	github.com/container-storage-interface/spec v1.8.0
 	github.com/golang/glog v0.0.0-20160126235308-23def4e6c14b
 	github.com/kubernetes-csi/csi-lib-utils v0.9.1
-	golang.org/x/net v0.0.0-20201202161906-c7110b5ffcbb
+	golang.org/x/net v0.7.0
 	golang.org/x/xerrors v0.0.0-20200804184101-5ec99f83aff1 // indirect
 	google.golang.org/grpc v1.33.1
 	google.golang.org/protobuf v1.24.0
@@ -41,8 +41,8 @@ github.com/chzyer/logex v1.1.10/go.mod h1:+Ywpsq7O8HXn0nuIou7OrIPyXbp3wmkHB+jjWR
 github.com/chzyer/readline v0.0.0-20180603132655-2972be24d48e/go.mod h1:nSuG5e5PlCu98SY8svDHJxuZscDgtXS6KTTbou5AhLI=
 github.com/chzyer/test v0.0.0-20180213035817-a1ea475d72b1/go.mod h1:Q3SI9o4m/ZMnBNeIyt5eFwwo7qiLfzFZmjNmxjkiQlU=
 github.com/container-storage-interface/spec v1.2.0/go.mod h1:6URME8mwIBbpVyZV93Ce5St17xBiQJQY67NDsuohiy4=
-github.com/container-storage-interface/spec v1.4.0 h1:ozAshSKxpJnYUfmkpZCTYyF/4MYeYlhdXbAvPvfGmkg=
-github.com/container-storage-interface/spec v1.4.0/go.mod h1:6URME8mwIBbpVyZV93Ce5St17xBiQJQY67NDsuohiy4=
+github.com/container-storage-interface/spec v1.8.0 h1:D0vhF3PLIZwlwZEf2eNbpujGCNwspwTYf2idJRJx4xI=
+github.com/container-storage-interface/spec v1.8.0/go.mod h1:ROLik+GhPslwwWRNFF1KasPzroNARibH2rfz1rkg4H0=
 github.com/davecgh/go-spew v1.1.0/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSsI+c5H38=
 github.com/davecgh/go-spew v1.1.1 h1:vj9j/u1bqnvCEfJOwUhtlOARqs3+rkHYY13jYWTU97c=
 github.com/davecgh/go-spew v1.1.1/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSsI+c5H38=
@@ -182,6 +182,7 @@ github.com/stretchr/testify v1.3.0/go.mod h1:M5WIy9Dh21IEIfnGCwXGc5bZfKNJtfHm1UV
 github.com/stretchr/testify v1.4.0/go.mod h1:j7eGeouHqKxXV5pUuKE4zz7dFj8WfuZ+81PSLYec5m4=
 github.com/stretchr/testify v1.5.1 h1:nOGnQDM7FYENwehXlg/kFVnos3rEvtKTjRvOWSzb6H4=
 github.com/stretchr/testify v1.5.1/go.mod h1:5W2xD1RspED5o8YsWQXVCued0rvSQ+mT+I5cxcmMvtA=
+github.com/yuin/goldmark v1.4.13/go.mod h1:6yULJ656Px+3vBD8DxQVa3kxgyrAnzto9xy5taEt/CY=
 go.opencensus.io v0.21.0/go.mod h1:mSImk1erAIZhrmZN+AvHh14ztQfjbGwt4TtuofqLduU=
 go.opencensus.io v0.22.0/go.mod h1:+kGneAE2xo2IficOXnaByMWTGM9T73dGwxeWcUqIpI8=
 go.opencensus.io v0.22.2/go.mod h1:yxeiOL68Rb0Xd1ddK5vPZ/oVn4vY4Ynel7k9FzqtOIw=
@@ -195,6 +196,7 @@ golang.org/x/crypto v0.0.0-20190605123033-f99c8df09eb5/go.mod h1:yigFU9vqHzYiE8U
 golang.org/x/crypto v0.0.0-20191011191535-87dc89f01550/go.mod h1:yigFU9vqHzYiE8UmvKecakEJjdnWj3jj499lnFckfCI=
 golang.org/x/crypto v0.0.0-20191206172530-e9b2fee46413/go.mod h1:LzIPMQfyMNhhGPhUkYOs5KpL4U8rLKemX1yGLhDgUto=
 golang.org/x/crypto v0.0.0-20200622213623-75b288015ac9/go.mod h1:LzIPMQfyMNhhGPhUkYOs5KpL4U8rLKemX1yGLhDgUto=
+golang.org/x/crypto v0.0.0-20210921155107-089bfa567519/go.mod h1:GvvjBRRGRdwPK5ydBHafDWAxML/pGHZbMvKqRZ5+Abc=
 golang.org/x/exp v0.0.0-20190121172915-509febef88a4/go.mod h1:CJ0aWSM057203Lf6IL+f9T1iT9GByDxfZKAQTCR3kQA=
 golang.org/x/exp v0.0.0-20190306152737-a1d7652674e8/go.mod h1:CJ0aWSM057203Lf6IL+f9T1iT9GByDxfZKAQTCR3kQA=
 golang.org/x/exp v0.0.0-20190510132918-efd6b22b2522/go.mod h1:ZjyILWgesfNpC6sMxTJOJm9Kp84zZh5NQWvqDGG3Qr8=
@@ -213,6 +215,7 @@ golang.org/x/mobile v0.0.0-20190719004257-d2bd2a29d028/go.mod h1:E/iHnbuqvinMTCc
 golang.org/x/mod v0.0.0-20190513183733-4bf6d317e70e/go.mod h1:mXi4GBBbnImb6dmsKGUJ2LatrhH/nqhxcFungHvyanc=
 golang.org/x/mod v0.1.0/go.mod h1:0QHyrYULN0/3qlju5TqG8bIK38QM8yzMo5ekMj3DlcY=
 golang.org/x/mod v0.1.1-0.20191105210325-c90efee705ee/go.mod h1:QqPTAvyqsEbceGzBzNggFXnrqF1CaUcvgkdR5Ot7KZg=
+golang.org/x/mod v0.6.0-dev.0.20220419223038-86c51ed26bb4/go.mod h1:jJ57K6gSWd91VN4djpZkiMVwK6gcyfeH4XE8wZrZaV4=
 golang.org/x/net v0.0.0-20180724234803-3673e40ba225/go.mod h1:mL1N/T3taQHkDXs73rZJwtUhF3w3ftmwwsq0BUmARs4=
 golang.org/x/net v0.0.0-20180906233101-161cd47e91fd/go.mod h1:mL1N/T3taQHkDXs73rZJwtUhF3w3ftmwwsq0BUmARs4=
 golang.org/x/net v0.0.0-20181114220301-adae6a3d119a/go.mod h1:mL1N/T3taQHkDXs73rZJwtUhF3w3ftmwwsq0BUmARs4=
@@ -228,8 +231,10 @@ golang.org/x/net v0.0.0-20190620200207-3b0461eec859/go.mod h1:z5CRVTTTmAJ677TzLL
 golang.org/x/net v0.0.0-20191209160850-c0dbc17a3553/go.mod h1:z5CRVTTTmAJ677TzLLGU+0bjPO0LkuOLi4/5GtJWs/s=
 golang.org/x/net v0.0.0-20200324143707-d3edc9973b7e/go.mod h1:qpuaurCH72eLCgpAm/N6yyVIVM9cpaDIP3A8BGJEC5A=
 golang.org/x/net v0.0.0-20200707034311-ab3426394381/go.mod h1:/O7V0waA8r7cgGh81Ro3o1hOxt32SMVPicZroKQ2sZA=
-golang.org/x/net v0.0.0-20201202161906-c7110b5ffcbb h1:eBmm0M9fYhWpKZLjQUUKka/LtIxf46G4fxeEz5KJr9U=
-golang.org/x/net v0.0.0-20201202161906-c7110b5ffcbb/go.mod h1:sp8m0HH+o8qH0wwXwYZr8TS3Oi6o0r6Gce1SSxlDquU=
+golang.org/x/net v0.0.0-20210226172049-e18ecbb05110/go.mod h1:m0MpNAwzfU5UDzcl9v0D8zg8gWTRqZa9RBIspLL5mdg=
+golang.org/x/net v0.0.0-20220722155237-a158d28d115b/go.mod h1:XRhObCWvk6IyKnWLug+ECip1KBveYUHfp+8e9klMJ9c=
+golang.org/x/net v0.7.0 h1:rJrUqqhjsgNp7KqAIc25s9pZnjU7TUcSY7HcVZjdn1g=
+golang.org/x/net v0.7.0/go.mod h1:2Tu9+aMcznHK/AK1HMvgo6xiTLG5rD5rZLDS+rp2Bjs=
 golang.org/x/oauth2 v0.0.0-20180821212333-d2e6202438be/go.mod h1:N/0e6XlmueqKjAGxoOufVs8QHGRruUQn6yWY3a++T0U=
 golang.org/x/oauth2 v0.0.0-20190226205417-e64efc72b421/go.mod h1:gOpvHmFTYa4IltrdGE7lF6nIHvwfUNPOp7c8zoXwtLw=
 golang.org/x/oauth2 v0.0.0-20190604053449-0f29369cfe45/go.mod h1:gOpvHmFTYa4IltrdGE7lF6nIHvwfUNPOp7c8zoXwtLw=
@@ -240,6 +245,7 @@ golang.org/x/sync v0.0.0-20181221193216-37e7f081c4d4/go.mod h1:RxMgew5VJxzue5/jJ
 golang.org/x/sync v0.0.0-20190227155943-e225da77a7e6/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM=
 golang.org/x/sync v0.0.0-20190423024810-112230192c58/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM=
 golang.org/x/sync v0.0.0-20190911185100-cd5d95a43a6e/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM=
+golang.org/x/sync v0.0.0-20220722155255-886fb9371eb4/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM=
 golang.org/x/sys v0.0.0-20180905080454-ebe1bf3edb33/go.mod h1:STP8DvDyc/dI5b8T5hshtkjS+E42TnysNCUPdjciGhY=
 golang.org/x/sys v0.0.0-20180909124046-d0be0721c37e/go.mod h1:STP8DvDyc/dI5b8T5hshtkjS+E42TnysNCUPdjciGhY=
 golang.org/x/sys v0.0.0-20181116152217-5ac8a444bdc5/go.mod h1:STP8DvDyc/dI5b8T5hshtkjS+E42TnysNCUPdjciGhY=
@@ -259,13 +265,22 @@ golang.org/x/sys v0.0.0-20200302150141-5c8b2ff67527/go.mod h1:h1NjWce9XRLGQEsW7w
 golang.org/x/sys v0.0.0-20200323222414-85ca7c5b95cd/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
 golang.org/x/sys v0.0.0-20200615200032-f1bc736245b1/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
 golang.org/x/sys v0.0.0-20200622214017-ed371f2e16b4/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
-golang.org/x/sys v0.0.0-20200930185726-fdedc70b468f h1:+Nyd8tzPX9R7BWHguqsrbFdRx3WQ/1ib8I44HXV5yTA=
-golang.org/x/sys v0.0.0-20200930185726-fdedc70b468f/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
+golang.org/x/sys v0.0.0-20201119102817-f84b799fce68/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
+golang.org/x/sys v0.0.0-20210615035016-665e8c7367d1/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
+golang.org/x/sys v0.0.0-20220520151302-bc2c85ada10a/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
+golang.org/x/sys v0.0.0-20220722155257-8c9f86f7a55f/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
+golang.org/x/sys v0.5.0 h1:MUK/U/4lj1t1oPg0HfuXDN/Z1wv31ZJ/YcPiGccS4DU=
+golang.org/x/sys v0.5.0/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
+golang.org/x/term v0.0.0-20201126162022-7de9c90e9dd1/go.mod h1:bj7SfCRtBDWHUb9snDiAeCFNEtKQo2Wmx5Cou7ajbmo=
+golang.org/x/term v0.0.0-20210927222741-03fcf44c2211/go.mod h1:jbD1KX2456YbFQfuXm/mYQcufACuNUgVhRMnK/tPxf8=
+golang.org/x/term v0.5.0/go.mod h1:jMB1sMXY+tzblOD4FWmEbocvup2/aLOaQEp7JmGp78k=
 golang.org/x/text v0.3.0/go.mod h1:NqM8EUOU14njkJ3fqMW+pc6Ldnwhi/IjpwHt7yyuwOQ=
 golang.org/x/text v0.3.1-0.20180807135948-17ff2d5776d2/go.mod h1:NqM8EUOU14njkJ3fqMW+pc6Ldnwhi/IjpwHt7yyuwOQ=
 golang.org/x/text v0.3.2/go.mod h1:bEr9sfX3Q8Zfm5fL9x+3itogRgK3+ptLWKqgva+5dAk=
-golang.org/x/text v0.3.3 h1:cokOdA+Jmi5PJGXLlLllQSgYigAEfHXJAERHVMaCc2k=
 golang.org/x/text v0.3.3/go.mod h1:5Zoc/QRtKVWzQhOtBMvqHzDpF6irO9z98xDceosuGiQ=
+golang.org/x/text v0.3.7/go.mod h1:u+2+/6zg+i71rQMx5EYifcz6MCKuco9NR6JIITiCfzQ=
+golang.org/x/text v0.7.0 h1:4BRB4x83lYWy72KwLD/qYDuTu7q9PjSagHvijDw7cLo=
+golang.org/x/text v0.7.0/go.mod h1:mrYo+phRRbMaCq/xk9113O4dZlRixOauAjOtrjsXDZ8=
 golang.org/x/time v0.0.0-20181108054448-85acf8d2951c/go.mod h1:tRJNPiyCQ0inRvYxbN9jk5I+vvW/OXSQhTDSoE431IQ=
 golang.org/x/time v0.0.0-20190308202827-9d24e82272b4/go.mod h1:tRJNPiyCQ0inRvYxbN9jk5I+vvW/OXSQhTDSoE431IQ=
 golang.org/x/time v0.0.0-20191024005414-555d28b269f0/go.mod h1:tRJNPiyCQ0inRvYxbN9jk5I+vvW/OXSQhTDSoE431IQ=
@@ -286,8 +301,10 @@ golang.org/x/tools v0.0.0-20190628153133-6cdbf07be9d0/go.mod h1:/rFqwRUd4F7ZHNgw
 golang.org/x/tools v0.0.0-20190816200558-6889da9d5479/go.mod h1:b+2E5dAYhXwXZwtnZ6UAqBI28+e2cm9otk0dWdXHAEo=
 golang.org/x/tools v0.0.0-20190911174233-4f2ddba30aff/go.mod h1:b+2E5dAYhXwXZwtnZ6UAqBI28+e2cm9otk0dWdXHAEo=
 golang.org/x/tools v0.0.0-20191012152004-8de300cfc20a/go.mod h1:b+2E5dAYhXwXZwtnZ6UAqBI28+e2cm9otk0dWdXHAEo=
+golang.org/x/tools v0.0.0-20191119224855-298f0cb1881e/go.mod h1:b+2E5dAYhXwXZwtnZ6UAqBI28+e2cm9otk0dWdXHAEo=
 golang.org/x/tools v0.0.0-20191125144606-a911d9008d1f/go.mod h1:b+2E5dAYhXwXZwtnZ6UAqBI28+e2cm9otk0dWdXHAEo=
 golang.org/x/tools v0.0.0-20191227053925-7b8e75db28f4/go.mod h1:TB2adYChydJhpapKDTa4BR/hXlZSLoq2Wpct/0txZ28=
+golang.org/x/tools v0.1.12/go.mod h1:hNGJHUnrk76NpqgfD5Aqm5Crs+Hm0VOH/i9J2+nxYbc=
 golang.org/x/xerrors v0.0.0-20190717185122-a985d3407aa7/go.mod h1:I/5z698sn9Ka8TeJc9MKroUUfqBBauWjQqLJ2OPfmY0=
 golang.org/x/xerrors v0.0.0-20191011141410-1b5146add898/go.mod h1:I/5z698sn9Ka8TeJc9MKroUUfqBBauWjQqLJ2OPfmY0=
 golang.org/x/xerrors v0.0.0-20191204190536-9bdfabe68543/go.mod h1:I/5z698sn9Ka8TeJc9MKroUUfqBBauWjQqLJ2OPfmY0=
@@ -5,7 +5,7 @@ package vitastor

 const (
    vitastorCSIDriverName    = "csi.vitastor.io"
-    vitastorCSIDriverVersion = "1.8.0"
+    vitastorCSIDriverVersion = "1.10.0"
 )

 // Config struct fills the parameters of request or user input
@@ -8,11 +8,8 @@ import (
    "encoding/json"
    "fmt"
    "strings"
-    "bytes"
-    "strconv"
    "time"
    "os"
-    "os/exec"
    "io/ioutil"

    "github.com/kubernetes-csi/csi-lib-utils/protosanitizer"
@@ -70,9 +67,10 @@ func GetConnectionParams(params map[string]string) (map[string]string, error)
    {
        configPath = "/etc/vitastor/vitastor.conf"
    }
-    else
+    ctxVars["configPath"] = configPath
+    if (params["vitastorfs"] != "")
    {
-        ctxVars["configPath"] = configPath
+        ctxVars["vitastorfs"] = params["vitastorfs"]
    }
    config := make(map[string]interface{})
    configFD, err := os.Open(configPath)
@@ -114,22 +112,6 @@ func GetConnectionParams(params map[string]string) (map[string]string, error)
    return ctxVars, nil
 }

-func system(program string, args ...string) ([]byte, []byte, error)
-{
-    klog.Infof("Running "+program+" "+strings.Join(args, " "))
-    c := exec.Command(program, args...)
-    var stdout, stderr bytes.Buffer
-    c.Stdout, c.Stderr = &stdout, &stderr
-    err := c.Run()
-    if (err != nil)
-    {
-        stdoutStr, stderrStr := string(stdout.Bytes()), string(stderr.Bytes())
-        klog.Errorf(program+" "+strings.Join(args, " ")+" failed: %s, status %s\n", stdoutStr+stderrStr, err)
-        return nil, nil, status.Error(codes.Internal, stdoutStr+stderrStr+" (status "+err.Error()+")")
-    }
-    return stdout.Bytes(), stderr.Bytes(), nil
-}
-
 func invokeCLI(ctxVars map[string]string, args []string) ([]byte, error)
 {
    if (ctxVars["configPath"] != "")
@@ -158,27 +140,57 @@ func (cs *ControllerServer) CreateVolume(ctx context.Context, req *csi.CreateVol
        return nil, status.Error(codes.InvalidArgument, "volume capabilities is a required field")
    }

-    etcdVolumePrefix := req.Parameters["etcdVolumePrefix"]
-    poolId, _ := strconv.ParseUint(req.Parameters["poolId"], 10, 64)
-    if (poolId == 0)
-    {
-        return nil, status.Error(codes.InvalidArgument, "poolId is missing in storage class configuration")
-    }
-
-    volName := etcdVolumePrefix + req.GetName()
-    volSize := 1 * GB
-    if capRange := req.GetCapacityRange(); capRange != nil
-    {
-        volSize = ((capRange.GetRequiredBytes() + MB - 1) / MB) * MB
-    }
-
    ctxVars, err := GetConnectionParams(req.Parameters)
    if (err != nil)
    {
        return nil, err
    }

-    args := []string{ "create", volName, "-s", fmt.Sprintf("%v", volSize), "--pool", fmt.Sprintf("%v", poolId) }
+    err = cs.checkCaps(volumeCapabilities, ctxVars["vitastorfs"] != "")
+    if (err != nil)
+    {
+        return nil, err
+    }
+
+    pool := req.Parameters["poolId"]
+    if (pool == "")
+    {
+        return nil, status.Error(codes.InvalidArgument, "poolId is missing in storage class configuration")
+    }
+    volumePrefix := req.Parameters["volumePrefix"]
+    if (volumePrefix == "")
+    {
+        // Old name
+        volumePrefix = req.Parameters["etcdVolumePrefix"]
+    }
+    volName := volumePrefix + req.GetName()
+    volSize := 1 * GB
+    if capRange := req.GetCapacityRange(); capRange != nil
+    {
+        volSize = ((capRange.GetRequiredBytes() + MB - 1) / MB) * MB
+    }
+
+    if (ctxVars["vitastorfs"] != "")
+    {
+        // Nothing to create, subdirectories are created during mounting
+        // FIXME: It would be cool to support quotas some day and set it here
+        if (req.VolumeContentSource.GetSnapshot() != nil)
+        {
+            return nil, status.Error(codes.InvalidArgument, "VitastorFS doesn't support snapshots")
+        }
+        ctxVars["name"] = volName
+        ctxVars["pool"] = pool
+        volumeIdJson, _ := json.Marshal(ctxVars)
+        return &csi.CreateVolumeResponse{
+            Volume: &csi.Volume{
+                // Ugly, but VolumeContext isn't passed to DeleteVolume :-(
+                VolumeId: string(volumeIdJson),
+                CapacityBytes: volSize,
+            },
+        }, nil
+    }
+
+    args := []string{ "create", volName, "-s", fmt.Sprintf("%v", volSize), "--pool", pool }

    // Support creation from snapshot
    var src *csi.VolumeContentSource
@@ -261,6 +273,12 @@ func (cs *ControllerServer) DeleteVolume(ctx context.Context, req *csi.DeleteVol
        return nil, err
    }

+    if (ctxVars["vitastorfs"] != "")
+    {
+        // FIXME: Delete FS subdirectory
+        return &csi.DeleteVolumeResponse{}, nil
+    }
+
    _, err = invokeCLI(ctxVars, []string{ "rm", volName })
    if (err != nil)
    {
@@ -295,19 +313,72 @@ func (cs *ControllerServer) ValidateVolumeCapabilities(ctx context.Context, req
    {
        return nil, status.Error(codes.InvalidArgument, "volumeId is nil")
    }
+    volVars := make(map[string]string)
+    err := json.Unmarshal([]byte(volumeID), &volVars)
+    if (err != nil)
+    {
+        return nil, status.Error(codes.Internal, "volume ID not in JSON format")
+    }
+    ctxVars, err := GetConnectionParams(volVars)
+    if (err != nil)
+    {
+        return nil, err
+    }
+
    volumeCapabilities := req.GetVolumeCapabilities()
    if (volumeCapabilities == nil)
    {
        return nil, status.Error(codes.InvalidArgument, "volumeCapabilities is nil")
    }

+    err = cs.checkCaps(volumeCapabilities, ctxVars["vitastorfs"] != "")
+    if (err != nil)
+    {
+        return nil, err
+    }
+
+    return &csi.ValidateVolumeCapabilitiesResponse{
+        Confirmed: &csi.ValidateVolumeCapabilitiesResponse_Confirmed{
+            VolumeCapabilities: req.VolumeCapabilities,
+        },
+    }, nil
+}
+
+func (cs *ControllerServer) checkCaps(volumeCapabilities []*csi.VolumeCapability, fs bool) error
+{
    var volumeCapabilityAccessModes []*csi.VolumeCapability_AccessMode
    for _, mode := range []csi.VolumeCapability_AccessMode_Mode{
        csi.VolumeCapability_AccessMode_SINGLE_NODE_WRITER,
-        csi.VolumeCapability_AccessMode_MULTI_NODE_MULTI_WRITER,
+        csi.VolumeCapability_AccessMode_SINGLE_NODE_READER_ONLY,
+        csi.VolumeCapability_AccessMode_MULTI_NODE_READER_ONLY,
+        csi.VolumeCapability_AccessMode_SINGLE_NODE_SINGLE_WRITER,
+        csi.VolumeCapability_AccessMode_SINGLE_NODE_MULTI_WRITER,
    } {
        volumeCapabilityAccessModes = append(volumeCapabilityAccessModes, &csi.VolumeCapability_AccessMode{Mode: mode})
    }
+    for _, capability := range volumeCapabilities
+    {
+        if (capability.GetBlock() != nil)
+        {
+            if (fs)
+            {
+                return status.Errorf(codes.InvalidArgument, "%v not supported with FS-based volumes", capability)
+            }
+            for _, mode := range []csi.VolumeCapability_AccessMode_Mode{
+                csi.VolumeCapability_AccessMode_MULTI_NODE_SINGLE_WRITER,
+                csi.VolumeCapability_AccessMode_MULTI_NODE_MULTI_WRITER,
+            } {
+                volumeCapabilityAccessModes = append(volumeCapabilityAccessModes, &csi.VolumeCapability_AccessMode{Mode: mode})
+            }
+            break
+        }
+    }
+
+    if (fs)
+    {
+        // All access modes including RWX are supported with FS-based volumes
+        return nil
+    }

    capabilitySupport := false
    for _, capability := range volumeCapabilities
@@ -323,14 +394,10 @@ func (cs *ControllerServer) ValidateVolumeCapabilities(ctx context.Context, req

    if (!capabilitySupport)
    {
-        return nil, status.Errorf(codes.NotFound, "%v not supported", req.GetVolumeCapabilities())
+        return status.Errorf(codes.InvalidArgument, "%v not supported", volumeCapabilities)
    }

-    return &csi.ValidateVolumeCapabilitiesResponse{
-        Confirmed: &csi.ValidateVolumeCapabilitiesResponse_Confirmed{
-            VolumeCapabilities: req.VolumeCapabilities,
-        },
-    }, nil
+    return nil
 }

 // ListVolumes returns a list of volumes
@@ -419,6 +486,12 @@ func (cs *ControllerServer) CreateSnapshot(ctx context.Context, req *csi.CreateS
    {
        return nil, status.Error(codes.Internal, "volume ID not in JSON format")
    }
+
+    if (ctxVars["vitastorfs"] != "")
+    {
+        return nil, status.Error(codes.InvalidArgument, "VitastorFS doesn't support snapshots")
+    }
+
    volName := ctxVars["name"]

    // Create image using vitastor-cli
@@ -477,6 +550,11 @@ func (cs *ControllerServer) DeleteSnapshot(ctx context.Context, req *csi.DeleteS
        return nil, err
    }

+    if (ctxVars["vitastorfs"] != "")
+    {
+        return nil, status.Error(codes.InvalidArgument, "VitastorFS doesn't support snapshots")
+    }
+
    _, err = invokeCLI(ctxVars, []string{ "rm", volName+"@"+snapName })
    if (err != nil)
    {
@@ -508,6 +586,11 @@ func (cs *ControllerServer) ListSnapshots(ctx context.Context, req *csi.ListSnap
        return nil, err
    }

+    if (ctxVars["vitastorfs"] != "")
+    {
+        return nil, status.Error(codes.InvalidArgument, "VitastorFS doesn't support snapshots")
+    }
+
    inodeCfg, err := invokeList(ctxVars, volName+"@*", false)
    if (err != nil)
    {
@@ -571,6 +654,16 @@ func (cs *ControllerServer) ControllerExpandVolume(ctx context.Context, req *csi
        return nil, err
    }

+    if (ctxVars["vitastorfs"] != "")
+    {
+        // Nothing to change
+        // FIXME: Support quotas and change quota here
+        return &csi.ControllerExpandVolumeResponse{
+            CapacityBytes: req.CapacityRange.RequiredBytes,
+            NodeExpansionRequired: false,
+        }, nil
+    }
+
    inodeCfg, err := invokeList(ctxVars, volName, true)
    if (err != nil)
    {
@@ -5,11 +5,15 @@ package vitastor

 import (
    "context"
+    "crypto/sha1"
+    "encoding/hex"
    "encoding/json"
    "fmt"
    "os"
    "os/exec"
    "path/filepath"
+    "regexp"
+    "strconv"
    "strings"
    "sync"
    "syscall"
@@ -29,13 +33,14 @@ import (
 type NodeServer struct
 {
    *Driver
-    useVduse bool
-    stateDir string
-    mounter mount.Interface
+    useVduse        bool
+    stateDir        string
+    nfsStageDir     string
+    mounter         mount.Interface
    restartInterval time.Duration
-    mu sync.Mutex
-    cond *sync.Cond
-    volumeLocks map[string]bool
+    mu              sync.Mutex
+    cond            *sync.Cond
+    volumeLocks     map[string]bool
 }

 type DeviceState struct
@@ -48,6 +53,15 @@ type DeviceState struct
    PidFile    string `json:"pidFile"`
 }

+type NfsState struct
+{
+    ConfigPath string `json:"configPath"`
+    FsName     string `json:"fsName"`
+    Pool       string `json:"pool"`
+    Path       string `json:"path"`
+    Port       int    `json:"port"`
+}
+
 // NewNodeServer create new instance node
 func NewNodeServer(driver *Driver) *NodeServer
 {
@@ -60,11 +74,17 @@ func NewNodeServer(driver *Driver) *NodeServer
    {
        stateDir += "/"
    }
+    nfsStageDir := os.Getenv("NFS_STAGE_DIR")
+    if (nfsStageDir == "")
+    {
+        nfsStageDir = "/var/lib/kubelet/plugins/csi.vitastor.io/nfs"
+    }
    ns := &NodeServer{
-        Driver: driver,
-        useVduse: checkVduseSupport(),
-        stateDir: stateDir,
-        mounter: mount.New(""),
+        Driver:      driver,
+        useVduse:    checkVduseSupport(),
+        stateDir:    stateDir,
+        nfsStageDir: nfsStageDir,
+        mounter:     mount.New(""),
        volumeLocks: make(map[string]bool),
    }
    ns.cond = sync.NewCond(&ns.mu)
@@ -123,12 +143,12 @@ func (ns *NodeServer) restarter()
 func (ns *NodeServer) restoreVduseDaemons()
 {
    pattern := ns.stateDir+"vitastor-vduse-*.json"
-    matches, err := filepath.Glob(pattern)
+    stateFiles, err := filepath.Glob(pattern)
    if (err != nil)
    {
        klog.Errorf("failed to list %s: %v", pattern, err)
    }
-    if (len(matches) == 0)
+    if (len(stateFiles) == 0)
    {
        return
    }
@@ -146,59 +166,162 @@ func (ns *NodeServer) restoreVduseDaemons()
        klog.Errorf("/sbin/vdpa -j dev list returned bad JSON (error %v): %v", err, string(devListJSON))
        return
    }
-    for _, stateFile := range matches
+    for _, stateFile := range stateFiles
    {
-        vdpaId := filepath.Base(stateFile)
-        vdpaId = vdpaId[0:len(vdpaId)-5]
-        // Check if VDPA device is still added to the bus
-        if (devs[vdpaId] == nil)
-        {
-            // Unused, clean it up
-            unmapVduseById(ns.stateDir, vdpaId)
-            continue
-        }
+        ns.checkVduseState(stateFile, devs)
+    }
+}

-        stateJSON, err := os.ReadFile(stateFile)
+func (ns *NodeServer) checkVduseState(stateFile string, devs map[string]interface{})
+{
+    // Check if VDPA device is still added to the bus
+    vdpaId := filepath.Base(stateFile)
+    vdpaId = vdpaId[0:len(vdpaId)-5]
+    if (devs[vdpaId] == nil)
+    {
+        // Unused, clean it up
+        unmapVduseById(ns.stateDir, vdpaId)
+        return
+    }
+
+    // Read state file
+    stateJSON, err := os.ReadFile(stateFile)
+    if (err != nil)
+    {
+        klog.Warningf("error reading state file %v: %v", stateFile, err)
+        return
+    }
+    var state DeviceState
+    err = json.Unmarshal(stateJSON, &state)
+    if (err != nil)
+    {
+        klog.Warningf("state file %v contains invalid JSON (error %v): %v", stateFile, err, string(stateJSON))
+        return
+    }
+
+    // Lock volume
+    ns.lockVolume(state.ConfigPath+":block:"+state.Image)
+    defer ns.unlockVolume(state.ConfigPath+":block:"+state.Image)
+
+    // Recheck state file after locking
+    _, err = os.ReadFile(stateFile)
+    if (err != nil)
+    {
+        klog.Warningf("state file %v disappeared, skipping volume", stateFile)
+        return
+    }
+
+    // Check if the storage daemon is still active
+    pidFile := ns.stateDir + vdpaId + ".pid"
+    exists := false
+    proc, err := findByPidFile(pidFile)
+    if (err == nil)
+    {
+        exists = proc.Signal(syscall.Signal(0)) == nil
+    }
+    if (!exists)
+    {
+        // Restart daemon
+        klog.Warningf("restarting storage daemon for volume %v (VDPA ID %v)", state.Image, vdpaId)
+        err = startStorageDaemon(vdpaId, state.Image, pidFile, state.ConfigPath, state.Readonly)
        if (err != nil)
        {
-            klog.Warningf("error reading state file %v: %v", stateFile, err)
-            continue
+            klog.Warningf("failed to restart storage daemon for volume %v: %v", state.Image, err)
        }
-        var state DeviceState
-        err = json.Unmarshal(stateJSON, &state)
+    }
+}
+
+func (ns *NodeServer) restoreNfsDaemons()
+{
+    pattern := ns.stateDir+"vitastor-nfs-*.json"
+    stateFiles, err := filepath.Glob(pattern)
+    if (err != nil)
+    {
+        klog.Errorf("failed to list %s: %v", pattern, err)
+    }
+    if (len(stateFiles) == 0)
+    {
+        return
+    }
+    activeNFS, err := ns.listActiveNFS()
+    if (err != nil)
+    {
+        return
+    }
+    // Check all state files and try to restore active mounts
+    for _, stateFile := range stateFiles
+    {
+        ns.checkNfsState(stateFile, activeNFS)
+    }
+}
+
+func (ns *NodeServer) readNfsState(stateFile string, allowNotExists bool) (*NfsState, error)
+{
+    stateJSON, err := os.ReadFile(stateFile)
+    if (err != nil)
+    {
+        if (allowNotExists && os.IsNotExist(err))
+        {
+            return nil, nil
+        }
+        klog.Warningf("error reading state file %v: %v", stateFile, err)
+        return nil, err
+    }
+    var state NfsState
+    err = json.Unmarshal(stateJSON, &state)
+    if (err != nil)
+    {
+        klog.Warningf("state file %v contains invalid JSON (error %v): %v", stateFile, err, string(stateJSON))
+        return nil, err
+    }
+    return &state, nil
+}
+
+func (ns *NodeServer) checkNfsState(stateFile string, activeNfs map[int][]string)
+{
+    // Read state file
+    state, err := ns.readNfsState(stateFile, false)
+    if (err != nil)
+    {
+        return
+    }
+    // Lock FS
+    ns.lockVolume(state.ConfigPath+":fs:"+state.FsName)
+    defer ns.unlockVolume(state.ConfigPath+":fs:"+state.FsName)
+    // Check if NFS at this port is still mounted
+    pidFile := ns.stateDir + filepath.Base(stateFile)
+    pidFile = pidFile[0:len(pidFile)-5] + ".pid"
+    if (len(activeNfs[state.Port]) == 0)
+    {
+        // this is a stale state file, remove it
+        klog.Warningf("state file %v contains stale mount at port %d, removing it", stateFile, state.Port)
+        ns.stopNFS(stateFile, pidFile)
+        return
+    }
+    // Check PID file
+    exists := false
+    proc, err := findByPidFile(pidFile)
+    if (err == nil)
+    {
+        exists = proc.Signal(syscall.Signal(0)) == nil
+    }
+    if (!exists)
+    {
+        // Restart vitastor-nfs server
+        klog.Warningf("restarting NFS server for FS %v at port %v", state.FsName, state.Port)
+        _, _, err := system(
+            "/usr/bin/vitastor-nfs", "start",
+            "--pidfile", pidFile,
+            "--bind", "127.0.0.1",
+            "--port", fmt.Sprintf("%d", state.Port),
+            "--fs", state.FsName,
+            "--pool", state.Pool,
+            "--portmap", "0",
+        )
        if (err != nil)
        {
-            klog.Warningf("state file %v contains invalid JSON (error %v): %v", stateFile, err, string(stateJSON))
-            continue
+            klog.Warningf("failed to restart NFS server for FS %v: %v", state.FsName, err)
        }
-
-        ns.lockVolume(state.ConfigPath+":"+state.Image)
-
-        // Recheck state file after locking
-        _, err = os.ReadFile(stateFile)
-        if (err != nil)
-        {
-            klog.Warningf("state file %v disappeared, skipping volume", stateFile)
-            ns.unlockVolume(state.ConfigPath+":"+state.Image)
-            continue
-        }
-
-        // Check if the storage daemon is still active
-        pidFile := ns.stateDir + vdpaId + ".pid"
-        exists := false
-        proc, err := findByPidFile(pidFile)
-        if (err == nil)
-        {
-            exists = proc.Signal(syscall.Signal(0)) == nil
-        }
-        if (!exists)
-        {
-            // Restart daemon
-            klog.Warningf("restarting storage daemon for volume %v (VDPA ID %v)", state.Image, vdpaId)
-            _ = startStorageDaemon(vdpaId, state.Image, pidFile, state.ConfigPath, state.Readonly)
-        }
-
-        ns.unlockVolume(state.ConfigPath+":"+state.Image)
    }
 }

@@ -220,14 +343,44 @@ func (ns *NodeServer) NodeStageVolume(ctx context.Context, req *csi.NodeStageVol
    }
    volName := ctxVars["name"]

-    ns.lockVolume(ctxVars["configPath"]+":"+volName)
-    defer ns.unlockVolume(ctxVars["configPath"]+":"+volName)
+    if (ctxVars["vitastorfs"] != "")
+    {
+        return &csi.NodeStageVolumeResponse{}, nil
+    }
+
+    ns.lockVolume(ctxVars["configPath"]+":block:"+volName)
+    defer ns.unlockVolume(ctxVars["configPath"]+":block:"+volName)

    targetPath := req.GetStagingTargetPath()
    isBlock := req.GetVolumeCapability().GetBlock() != nil

    // Check that it's not already mounted
-    _, err = mount.IsNotMountPoint(ns.mounter, targetPath)
+    notmnt, err := mount.IsNotMountPoint(ns.mounter, targetPath)
+    if (err == nil)
+    {
+        if (!notmnt)
+        {
+            klog.Errorf("target path %s is already mounted", targetPath)
+            return nil, fmt.Errorf("target path %s is already mounted", targetPath)
+        }
+        var finfo os.FileInfo
+        finfo, err = os.Stat(targetPath)
+        if (err != nil)
+        {
+            klog.Errorf("failed to stat %s: %v", targetPath, err)
+            return nil, err
+        }
+        if (finfo.IsDir() != (!isBlock))
+        {
+            err = os.Remove(targetPath)
+            if (err != nil)
+            {
+                klog.Errorf("failed to remove %s (to recreate it with correct type): %v", targetPath, err)
+                return nil, err
+            }
+            err = os.ErrNotExist
+        }
+    }
    if (err != nil)
    {
        if (os.IsNotExist(err))
@@ -280,6 +433,7 @@ func (ns *NodeServer) NodeStageVolume(ctx context.Context, req *csi.NodeStageVol
    diskMounter := &mount.SafeFormatAndMount{Interface: ns.mounter, Exec: utilexec.New()}
    if (isBlock)
    {
+        klog.Infof("bind-mounting %s to %s", devicePath, targetPath)
        err = diskMounter.Mount(devicePath, targetPath, "", []string{"bind"})
    }
    else
@@ -309,39 +463,40 @@ func (ns *NodeServer) NodeStageVolume(ctx context.Context, req *csi.NodeStageVol
        readOnly := Contains(opt, "ro")
        if (existingFormat == "" && !readOnly)
        {
-            var cmdOut []byte
            switch fsType
            {
                case "ext4":
                    args := []string{"-m0", "-Enodiscard,lazy_itable_init=1,lazy_journal_init=1", devicePath}
-                    cmdOut, err = diskMounter.Exec.Command("mkfs.ext4", args...).CombinedOutput()
+                    _, err = systemCombined("mkfs.ext4", args...)
                case "xfs":
-                    cmdOut, err = diskMounter.Exec.Command("mkfs.xfs", "-K", devicePath).CombinedOutput()
+                    _, err = systemCombined("mkfs.xfs", "-K", devicePath)
            }
            if (err != nil)
            {
-                klog.Errorf("failed to run mkfs error: %v, output: %v", err, string(cmdOut))
                goto unmap
            }
        }

+        klog.Infof("formatting and mounting %s to %s with FS %s, options: %v", devicePath, targetPath, fsType, opt)
        err = diskMounter.FormatAndMount(devicePath, targetPath, fsType, opt)
+        if (err == nil)
+        {
+            klog.Infof("successfully mounted %s to %s", devicePath, targetPath)
+        }

        // Try to run online resize on mount.
        // FIXME: Implement online resize. It requires online resize support in vitastor-nbd.
        if (err == nil && existingFormat != "" && !readOnly)
        {
-            var cmdOut []byte
            switch (fsType)
            {
                case "ext4":
-                    cmdOut, err = diskMounter.Exec.Command("resize2fs", devicePath).CombinedOutput()
+                    _, err = systemCombined("resize2fs", devicePath)
                case "xfs":
-                    cmdOut, err = diskMounter.Exec.Command("xfs_growfs", devicePath).CombinedOutput()
+                    _, err = systemCombined("xfs_growfs", devicePath)
            }
            if (err != nil)
            {
-                klog.Errorf("failed to run resizefs error: %v, output: %v", err, string(cmdOut))
                goto unmap
            }
        }
@@ -381,11 +536,16 @@ func (ns *NodeServer) NodeUnstageVolume(ctx context.Context, req *csi.NodeUnstag
    }
    volName := ctxVars["name"]

-    ns.lockVolume(ctxVars["configPath"]+":"+volName)
-    defer ns.unlockVolume(ctxVars["configPath"]+":"+volName)
+    if (ctxVars["vitastorfs"] != "")
+    {
+        return &csi.NodeUnstageVolumeResponse{}, nil
+    }
+
+    ns.lockVolume(ctxVars["configPath"]+":block:"+volName)
+    defer ns.unlockVolume(ctxVars["configPath"]+":block:"+volName)

    targetPath := req.GetStagingTargetPath()
-    devicePath, refCount, err := mount.GetDeviceNameFromMount(ns.mounter, targetPath)
+    devicePath, _, err := mount.GetDeviceNameFromMount(ns.mounter, targetPath)
    if (err != nil)
    {
        if (os.IsNotExist(err))
@@ -402,6 +562,16 @@ func (ns *NodeServer) NodeUnstageVolume(ctx context.Context, req *csi.NodeUnstag
        return &csi.NodeUnstageVolumeResponse{}, nil
    }

+    refList, err := ns.mounter.GetMountRefs(targetPath)
+    if (err != nil)
+    {
+        return nil, err
+    }
+    if (len(refList) > 0)
+    {
+        klog.Warningf("%s is still referenced: %v", targetPath, refList)
+    }
+
    // unmount
    err = mount.CleanupMountPoint(targetPath, ns.mounter, false)
    if (err != nil)
@@ -410,7 +580,7 @@ func (ns *NodeServer) NodeUnstageVolume(ctx context.Context, req *csi.NodeUnstag
    }

    // unmap device
-    if (refCount == 1)
+    if (len(refList) == 0)
    {
        if (!ns.useVduse)
        {
@@ -425,6 +595,153 @@ func (ns *NodeServer) NodeUnstageVolume(ctx context.Context, req *csi.NodeUnstag
    return &csi.NodeUnstageVolumeResponse{}, nil
 }

+// Mount or check if NFS is already mounted
+func (ns *NodeServer) mountNFS(ctxVars map[string]string) (string, error)
+{
+    sum := sha1.Sum([]byte(ctxVars["configPath"]+":fs:"+ctxVars["vitastorfs"]))
+    nfsHash := hex.EncodeToString(sum[:])
+    stateFile := ns.stateDir+"vitastor-nfs-"+nfsHash+".json"
+    pidFile := ns.stateDir+"vitastor-nfs-"+nfsHash+".pid"
+    mountPath := ns.nfsStageDir+"/"+nfsHash
+    state, err := ns.readNfsState(stateFile, true)
+    if (state != nil)
+    {
+        return state.Path, nil
+    }
+    if (err != nil)
+    {
+        return "", err
+    }
+    err = os.MkdirAll(mountPath, 0777)
+    if (err != nil)
+    {
+        return "", err
+    }
+    // Create a new mount
+    state = &NfsState{
+        ConfigPath: ctxVars["configPath"],
+        FsName:     ctxVars["vitastorfs"],
+        Pool:       ctxVars["pool"],
+        Path:       mountPath,
+    }
+    klog.Infof("starting new NFS server for FS %v", state.FsName)
+    stdout, _, err := system(
+        "/usr/bin/vitastor-nfs", "start",
+        "--pidfile", pidFile,
+        "--bind", "127.0.0.1",
+        "--port", "auto",
+        "--fs", state.FsName,
+        "--pool", state.Pool,
+        "--portmap", "0",
+    )
+    if (err != nil)
+    {
+        return "", err
+    }
+    match := regexp.MustCompile("Port: (\\d+)").FindStringSubmatch(string(stdout))
+    if (match == nil)
+    {
+        klog.Errorf("failed to find port in vitastor-nfs output: %v", string(stdout))
+        ns.stopNFS(stateFile, pidFile)
+        return "", fmt.Errorf("failed to find port in vitastor-nfs output (bad vitastor-nfs version?)")
+    }
+    port, _ := strconv.ParseUint(match[1], 0, 16)
+    state.Port = int(port)
+    // Write state file
+    stateJSON, _ := json.Marshal(state)
+    err = os.WriteFile(stateFile, stateJSON, 0600)
+    if (err != nil)
+    {
+        klog.Errorf("failed to write state file %v", stateFile)
+        ns.stopNFS(stateFile, pidFile)
+        return "", err
+    }
+    // Mount NFS
+    _, _, err = system(
+        "mount", "-t", "nfs", "127.0.0.1:/", state.Path,
+        "-o", fmt.Sprintf("port=%d,mountport=%d,nfsvers=3,soft,nolock,tcp", port, port),
+    )
+    if (err != nil)
+    {
+        ns.stopNFS(stateFile, pidFile)
+        return "", err
+    }
+    return state.Path, nil
+}
+
+// Mount or check if NFS is already mounted
+func (ns *NodeServer) checkStopNFS(ctxVars map[string]string)
+{
+    sum := sha1.Sum([]byte(ctxVars["configPath"]+":fs:"+ctxVars["vitastorfs"]))
+    nfsHash := hex.EncodeToString(sum[:])
+    stateFile := ns.stateDir+"vitastor-nfs-"+nfsHash+".json"
+    pidFile := ns.stateDir+"vitastor-nfs-"+nfsHash+".pid"
+    mountPath := ns.nfsStageDir+"/"+nfsHash
+    state, err := ns.readNfsState(stateFile, true)
+    if (state == nil)
+    {
+        return
+    }
+    activeNFS, err := ns.listActiveNFS()
+    if (err != nil)
+    {
+        return
+    }
+    if (len(activeNFS[state.Port]) > 0)
+    {
+        return
+    }
+    // All volume mounts are detached, unmount the root mount and kill the server
+    err = mount.CleanupMountPoint(mountPath, ns.mounter, false)
+    if (err != nil)
+    {
+        klog.Errorf("failed to unmount %v: %v", mountPath, err)
+        return
+    }
+    ns.stopNFS(stateFile, pidFile)
+}
+
+func (ns *NodeServer) stopNFS(stateFile, pidFile string)
+{
+    err := killByPidFile(pidFile)
+    if (err != nil)
+    {
+        klog.Errorf("failed to kill process with pid from %v: %v", pidFile, err)
+    }
+    os.Remove(pidFile)
+    os.Remove(stateFile)
+}
+
+func (ns *NodeServer) listActiveNFS() (map[int][]string, error)
+{
+    mounts, err := mount.ParseMountInfo("/proc/self/mountinfo")
+    if (err != nil)
+    {
+        klog.Errorf("failed to list mounts: %v", err)
+        return nil, err
+    }
+    activeNFS := make(map[int][]string)
+    for _, mount := range mounts
+    {
+        // Volume mounts always refer to subpaths
+        if (mount.FsType == "nfs" && mount.Root != "/")
+        {
+            for _, opt := range mount.MountOptions
+            {
+                if (strings.HasPrefix(opt, "port="))
+                {
+                    port64, err := strconv.ParseUint(opt[5:], 10, 16)
+                    if (err == nil)
+                    {
+                        activeNFS[int(port64)] = append(activeNFS[int(port64)], mount.MountPoint)
+                    }
+                }
+            }
+        }
+    }
+    return activeNFS, nil
+}
+
 // NodePublishVolume mounts the volume mounted to the staging path to the target path
 func (ns *NodeServer) NodePublishVolume(ctx context.Context, req *csi.NodePublishVolumeRequest) (*csi.NodePublishVolumeResponse, error)
 {
@@ -443,23 +760,39 @@ func (ns *NodeServer) NodePublishVolume(ctx context.Context, req *csi.NodePublis
    }
    volName := ctxVars["name"]

-    ns.lockVolume(ctxVars["configPath"]+":"+volName)
-    defer ns.unlockVolume(ctxVars["configPath"]+":"+volName)
+    if (ctxVars["vitastorfs"] != "")
+    {
+        ns.lockVolume(ctxVars["configPath"]+":fs:"+ctxVars["vitastorfs"])
+        defer ns.unlockVolume(ctxVars["configPath"]+":fs:"+ctxVars["vitastorfs"])
+    }
+    else
+    {
+        ns.lockVolume(ctxVars["configPath"]+":block:"+volName)
+        defer ns.unlockVolume(ctxVars["configPath"]+":block:"+volName)
+    }

    stagingTargetPath := req.GetStagingTargetPath()
    targetPath := req.GetTargetPath()
    isBlock := req.GetVolumeCapability().GetBlock() != nil

-    // Check that stagingTargetPath is mounted
-    _, err = mount.IsNotMountPoint(ns.mounter, stagingTargetPath)
-    if (err != nil)
+    if (ctxVars["vitastorfs"] == "")
    {
-        klog.Errorf("staging path %v is not mounted: %v", stagingTargetPath, err)
-        return nil, fmt.Errorf("staging path %v is not mounted: %v", stagingTargetPath, err)
+        // Check that stagingTargetPath is mounted
+        notmnt, err := mount.IsNotMountPoint(ns.mounter, stagingTargetPath)
+        if (err != nil)
+        {
+            klog.Errorf("staging path %v is not mounted: %w", stagingTargetPath, err)
+            return nil, fmt.Errorf("staging path %v is not mounted: %w", stagingTargetPath, err)
+        }
+        else if (notmnt)
+        {
+            klog.Errorf("staging path %v is not mounted", stagingTargetPath)
+            return nil, fmt.Errorf("staging path %v is not mounted", stagingTargetPath)
+        }
    }

    // Check that targetPath is not already mounted
-    _, err = mount.IsNotMountPoint(ns.mounter, targetPath)
+    notmnt, err := mount.IsNotMountPoint(ns.mounter, targetPath)
    if (err != nil)
    {
        if (os.IsNotExist(err))
@@ -494,6 +827,29 @@ func (ns *NodeServer) NodePublishVolume(ctx context.Context, req *csi.NodePublis
            return nil, err
        }
    }
+    else if (!notmnt)
+    {
+        klog.Errorf("target path %s is already mounted", targetPath)
+        return nil, fmt.Errorf("target path %s is already mounted", targetPath)
+    }
+
+    if (ctxVars["vitastorfs"] != "")
+    {
+        nfspath, err := ns.mountNFS(ctxVars)
+        if (err != nil)
+        {
+            ns.checkStopNFS(ctxVars)
+            return nil, err
+        }
+        // volName should include prefix
+        stagingTargetPath = nfspath+"/"+volName
+        err = os.MkdirAll(stagingTargetPath, 0777)
+        if (err != nil && !os.IsExist(err))
+        {
+            ns.checkStopNFS(ctxVars)
+            return nil, err
+        }
+    }

    execArgs := []string{"--bind", stagingTargetPath, targetPath}
    if (req.GetReadonly())
@@ -506,6 +862,10 @@ func (ns *NodeServer) NodePublishVolume(ctx context.Context, req *csi.NodePublis
    out, err := cmd.Output()
    if (err != nil)
    {
+        if (ctxVars["vitastorfs"] != "")
+        {
+            ns.checkStopNFS(ctxVars)
+        }
        return nil, fmt.Errorf("Error running mount %v: %s", strings.Join(execArgs, " "), out)
    }

@@ -525,8 +885,16 @@ func (ns *NodeServer) NodeUnpublishVolume(ctx context.Context, req *csi.NodeUnpu
    }
    volName := ctxVars["name"]

-    ns.lockVolume(ctxVars["configPath"]+":"+volName)
-    defer ns.unlockVolume(ctxVars["configPath"]+":"+volName)
+    if (ctxVars["vitastorfs"] != "")
+    {
+        ns.lockVolume(ctxVars["configPath"]+":fs:"+ctxVars["vitastorfs"])
+        defer ns.unlockVolume(ctxVars["configPath"]+":fs:"+ctxVars["vitastorfs"])
+    }
+    else
+    {
+        ns.lockVolume(ctxVars["configPath"]+":block:"+volName)
+        defer ns.unlockVolume(ctxVars["configPath"]+":block:"+volName)
+    }

    targetPath := req.GetTargetPath()
    devicePath, _, err := mount.GetDeviceNameFromMount(ns.mounter, targetPath)
@@ -553,6 +921,11 @@ func (ns *NodeServer) NodeUnpublishVolume(ctx context.Context, req *csi.NodeUnpu
        return nil, err
    }

+    if (ctxVars["vitastorfs"] != "")
+    {
+        ns.checkStopNFS(ctxVars)
+    }
+
    return &csi.NodeUnpublishVolumeResponse{}, nil
 }

@@ -4,6 +4,7 @@
 package vitastor

 import (
+    "bytes"
    "errors"
    "encoding/json"
    "fmt"
@@ -15,6 +16,8 @@ import (
    "syscall"

    "k8s.io/klog"
+    "google.golang.org/grpc/codes"
+    "google.golang.org/grpc/status"
 )

 func Contains(list []string, s string) bool
@@ -73,6 +76,10 @@ func checkVduseSupport() bool
            " For VDUSE you need at least Linux 5.15 and the following kernel modules: vdpa, virtio-vdpa, vduse.",
        )
    }
+    else
+    {
+        klog.Infof("VDUSE support enabled successfully")
+    }
    return vduse
 }

@@ -97,6 +104,7 @@ func mapNbd(volName string, ctxVars map[string]string, readonly bool) (string, e
    {
        return "", fmt.Errorf("vitastor-nbd did not return the name of NBD device. output: %s", stderr)
    }
+    klog.Infof("Attached volume %s via NBD as %s", volName, dev)
    return dev, err
 }

@@ -217,6 +225,7 @@ func mapVduse(stateDir string, volName string, ctxVars map[string]string, readon
                    err = os.WriteFile(stateFile, stateJSON, 0600)
                    if (err == nil)
                    {
+                        klog.Infof("Attached volume %s via VDUSE as %s (VDPA ID %s)", volName, blockdev, vdpaId)
                        return blockdev, vdpaId, nil
                    }
                }
@@ -299,3 +308,35 @@ func unmapVduseById(stateDir, vdpaId string)
        os.Remove(pidFile)
    }
 }
+
+func system(program string, args ...string) ([]byte, []byte, error)
+{
+    klog.Infof("Running "+program+" "+strings.Join(args, " "))
+    c := exec.Command(program, args...)
+    var stdout, stderr bytes.Buffer
+    c.Stdout, c.Stderr = &stdout, &stderr
+    err := c.Run()
+    if (err != nil)
+    {
+        stdoutStr, stderrStr := string(stdout.Bytes()), string(stderr.Bytes())
+        klog.Errorf(program+" "+strings.Join(args, " ")+" failed: %s\nOutput:\n%s", err, stdoutStr+stderrStr)
+        return nil, nil, status.Error(codes.Internal, stdoutStr+stderrStr+" (status "+err.Error()+")")
+    }
+    return stdout.Bytes(), stderr.Bytes(), nil
+}
+
+func systemCombined(program string, args ...string) ([]byte, error)
+{
+    klog.Infof("Running "+program+" "+strings.Join(args, " "))
+    c := exec.Command(program, args...)
+    var out bytes.Buffer
+    c.Stdout, c.Stderr = &out, &out
+    err := c.Run()
+    if (err != nil)
+    {
+        outStr := string(out.Bytes())
+        klog.Errorf(program+" "+strings.Join(args, " ")+" failed: %s, status %s\n", outStr, err)
+        return nil, status.Error(codes.Internal, outStr+" (status "+err.Error()+")")
+    }
+    return out.Bytes(), nil
+}
@@ -1,4 +1,4 @@
-vitastor (1.8.0-1) unstable; urgency=medium
+vitastor (1.10.0-1) unstable; urgency=medium

  * Bugfixes

@@ -21,10 +21,10 @@ RUN set -e -x; \
    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 fio liburing-dev libgoogle-perftools-dev devscripts libjerasure-dev cmake libibverbs-dev libisal-dev libnl-3-dev libnl-genl-3-dev curl
-RUN apt-get -y build-dep fio
-RUN apt-get --download-only source fio
+RUN apt-get update && \
+    apt-get -y install fio liburing-dev libgoogle-perftools-dev devscripts libjerasure-dev cmake libibverbs-dev librdmacm-dev libisal-dev libnl-3-dev libnl-genl-3-dev curl && \
+    apt-get -y build-dep fio && \
+    apt-get --download-only source fio

 ADD . /root/vitastor
 RUN set -e -x; \
@@ -106,8 +106,8 @@ SSD cache or "media-cache" - for example, a lot of Seagate EXOS drives have
 it (they have internal SSD cache even though it's not stated in datasheets).

 Setting this parameter to "all" or "small" in OSD parameters requires enabling
-[disable_journal_fsync](layout-osd.en.yml#disable_journal_fsync) and
-[disable_meta_fsync](layout-osd.en.yml#disable_meta_fsync), setting it to
-"all" also requires enabling [disable_data_fsync](layout-osd.en.yml#disable_data_fsync).
+[disable_journal_fsync](layout-osd.en.md#disable_journal_fsync) and
+[disable_meta_fsync](layout-osd.en.md#disable_meta_fsync), setting it to
+"all" also requires enabling [disable_data_fsync](layout-osd.en.md#disable_data_fsync).
 vitastor-disk tried to do that by default, first checking/disabling drive cache.
 If it can't disable drive cache, OSD get initialized with "none".
@@ -112,6 +112,6 @@ HDD-дисках с внутренним SSD или "медиа" кэшем - н
 указано в спецификациях).

 Указание "all" или "small" в настройках / командной строке OSD требует
-включения [disable_journal_fsync](layout-osd.ru.yml#disable_journal_fsync) и
-[disable_meta_fsync](layout-osd.ru.yml#disable_meta_fsync), значение "all"
-также требует включения [disable_data_fsync](layout-osd.ru.yml#disable_data_fsync).
+включения [disable_journal_fsync](layout-osd.ru.md#disable_journal_fsync) и
+[disable_meta_fsync](layout-osd.ru.md#disable_meta_fsync), значение "all"
+также требует включения [disable_data_fsync](layout-osd.ru.md#disable_data_fsync).
@@ -118,12 +118,13 @@ Physical block size of the journal device. Must be a multiple of
 - Type: boolean
 - Default: false

-Do not issue fsyncs to the data device, i.e. do not flush its cache.
-Safe ONLY if your data device has write-through cache. If you disable
-the cache yourself using `hdparm` or `scsi_disk/cache_type` then make sure
-that the cache disable command is run every time before starting Vitastor
-OSD, for example, in the systemd unit. See also `immediate_commit` option
-for the instructions to disable cache and how to benefit from it.
+Do not issue fsyncs to the data device, i.e. do not force it to flush cache.
+Safe ONLY if your data device has write-through cache or if write-back
+cache is disabled. If you disable drive cache manually with `hdparm` or
+writing to `/sys/.../scsi_disk/cache_type` then make sure that you do it
+every time before starting Vitastor OSD (vitastor-disk does it automatically).
+See also [immediate_commit](layout-cluster.en.md#immediate_commit)
+for information about how to benefit from disabled cache.

 ## disable_meta_fsync

@@ -171,8 +172,7 @@ size, it actually has to write the whole 4 KB sector.

 Because of this it can actually be beneficial to use SSDs which work well
 with 512 byte sectors and use 512 byte disk_alignment, journal_block_size
-and meta_block_size. But the only SSD that may fit into this category is
-Intel Optane (probably, not tested yet).
+and meta_block_size. But at the moment, no such SSDs are known...

 Clients don't need to be aware of disk_alignment, so it's not required to
 put a modified value into etcd key /vitastor/config/global.
@@ -122,13 +122,14 @@ SSD-диске, иначе производительность пострада
 - Тип: булево (да/нет)
 - Значение по умолчанию: false

-Не отправлять fsync-и устройству данных, т.е. не сбрасывать его кэш.
+Не отправлять fsync-и устройству данных, т.е. не заставлять его сбрасывать кэш.
 Безопасно, ТОЛЬКО если ваше устройство данных имеет кэш со сквозной
-записью (write-through). Если вы отключаете кэш через `hdparm` или
-`scsi_disk/cache_type`, то удостоверьтесь, что команда отключения кэша
-выполняется перед каждым запуском Vitastor OSD, например, в systemd unit-е.
-Смотрите также опцию `immediate_commit` для инструкций по отключению кэша
-и о том, как из этого извлечь выгоду.
+записью (write-through) или если кэш с отложенной записью (write-back) отключён.
+Если вы отключаете кэш вручную через `hdparm` или запись в `/sys/.../scsi_disk/cache_type`,
+то удостоверьтесь, что вы делаете это каждый раз перед запуском Vitastor OSD
+(vitastor-disk делает это автоматически). Смотрите также опцию
+[immediate_commit](layout-cluster.ru.md#immediate_commit) для информации о том,
+как извлечь выгоду из отключённого кэша.

 ## disable_meta_fsync

@@ -179,9 +180,8 @@ SSD и HDD диски используют 4 КБ физические сект

 Поэтому, на самом деле, может быть выгодно найти SSD, хорошо работающие с
 меньшими, 512-байтными, блоками и использовать 512-байтные disk_alignment,
-journal_block_size и meta_block_size. Однако единственные SSD, которые
-теоретически могут попасть в эту категорию - это Intel Optane (но и это
-пока не проверялось автором).
+journal_block_size и meta_block_size. Однако на данный момент такие SSD
+не известны...

 Клиентам не обязательно знать про disk_alignment, так что помещать значение
 этого параметра в etcd в /vitastor/config/global не нужно.
@@ -24,6 +24,7 @@ These parameters only apply to Monitors.
 - [osd_out_time](#osd_out_time)
 - [placement_levels](#placement_levels)
 - [use_old_pg_combinator](#use_old_pg_combinator)
+- [osd_backfillfull_ratio](#osd_backfillfull_ratio)

 ## use_antietcd

@@ -175,3 +176,18 @@ present in the configuration, then it is defined with the default priority

 Use the old PG combination generator which doesn't support [level_placement](pool.en.md#level_placement)
 and [raw_placement](pool.en.md#raw_placement) for pools which don't use this features.
+
+## osd_backfillfull_ratio
+
+- Type: number
+- Default: 0.99
+
+Monitors try to prevent OSDs becoming 100% full during rebalance or recovery by
+calculating how much space will be occupied on every OSD after all rebalance
+and recovery operations finish, and pausing rebalance and recovery if that
+amount of space exceeds OSD capacity multiplied by the value of this
+configuration parameter.
+
+Future used space is calculated by summing space used by all user data blocks
+(objects) in all PGs placed on a specific OSD, even if some of these objects
+currently reside on a different set of OSDs.
@@ -24,6 +24,7 @@
 - [osd_out_time](#osd_out_time)
 - [placement_levels](#placement_levels)
 - [use_old_pg_combinator](#use_old_pg_combinator)
+- [osd_backfillfull_ratio](#osd_backfillfull_ratio)

 ## use_antietcd

@@ -178,3 +179,19 @@ OSD перед обновлением агрегированной статис

 Использовать старый генератор комбинаций PG, не поддерживающий [level_placement](pool.ru.md#level_placement)
 и [raw_placement](pool.ru.md#raw_placement) для пулов, которые не используют данные функции.
+
+## osd_backfillfull_ratio
+
+- Тип: число
+- Значение по умолчанию: 0.99
+
+Мониторы стараются предотвратить 100% заполнение OSD в процессе ребаланса
+или восстановления, рассчитывая, сколько места будет занято на каждом OSD после
+завершения всех операций ребаланса и восстановления, и приостанавливая
+ребаланс и восстановление, если рассчитанный объём превышает ёмкость OSD,
+умноженную на значение данного параметра.
+
+Будущее занятое место рассчитывается сложением места, занятого всеми
+пользовательскими блоками данных (объектами) во всех PG, расположенных
+на конкретном OSD, даже если часть этих объектов в данный момент находится
+на другом наборе OSD.
@@ -68,11 +68,17 @@ but they are not connected to the cluster.
 - Type: string

 RDMA device name to use for Vitastor OSD communications (for example,
-"rocep5s0f0"). Now Vitastor supports all adapters, even ones without
-ODP support, like Mellanox ConnectX-3 and non-Mellanox cards.
+"rocep5s0f0"). If not specified, Vitastor will try to find an RoCE
+device matching [osd_network](osd.en.md#osd_network), preferring RoCEv2,
+or choose the first available RDMA device if no RoCE devices are
+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.

-Versions up to Vitastor 1.2.0 required ODP which is only present in
-Mellanox ConnectX >= 4. See also [rdma_odp](#rdma_odp).
+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.
@@ -95,15 +101,17 @@ your device has.
 ## rdma_gid_index

 - Type: integer
- Default: 0

 Global address identifier index of the RDMA device to use. Different GID
 indexes may correspond to different protocols like RoCEv1, RoCEv2 and iWARP.
 Search for "GID" in `ibv_devinfo -v` output to determine which GID index
 you need.

-**IMPORTANT:** If you want to use RoCEv2 (as recommended) then the correct
-rdma_gid_index is usually 1 (IPv6) or 3 (IPv4).
+If not specified, Vitastor will try to auto-select a RoCEv2 IPv4 GID, then
+RoCEv2 IPv6 GID, then RoCEv1 IPv4 GID, then RoCEv1 IPv6 GID, then IB GID.
+GID auto-selection is unsupported with libibverbs < v32.
+
+A correct rdma_gid_index for RoCEv2 is usually 1 (IPv6) or 3 (IPv4).

 ## rdma_mtu

@@ -71,12 +71,17 @@ RDMA может быть нужно только если у клиентов е
 - Тип: строка

 Название RDMA-устройства для связи с Vitastor OSD (например, "rocep5s0f0").
-Сейчас 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.
+Если не указано, Vitastor попробует найти RoCE-устройство, соответствующее
+[osd_network](osd.en.md#osd_network), предпочитая RoCEv2, или выбрать первое
+попавшееся RDMA-устройство, если RoCE-устройств нет или если сеть `osd_network`
+не задана. Также автовыбор не поддерживается со старыми версиями библиотеки
+libibverbs < v32, например в Debian 10 Buster или CentOS 7.

-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).
+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.
@@ -101,15 +106,18 @@ Control) и ECN (Explicit Congestion Notification).
 ## rdma_gid_index

 - Тип: целое число
- Значение по умолчанию: 0

 Номер глобального идентификатора адреса RDMA-устройства, который следует
 использовать. Разным gid_index могут соответствовать разные протоколы связи:
 RoCEv1, RoCEv2, iWARP. Чтобы понять, какой нужен вам - смотрите строчки со
 словом "GID" в выводе команды `ibv_devinfo -v`.

-**ВАЖНО:** Если вы хотите использовать RoCEv2 (как мы и рекомендуем), то
-правильный rdma_gid_index, как правило, 1 (IPv6) или 3 (IPv4).
+Если не указан, Vitastor попробует автоматически выбрать сначала GID,
+соответствующий RoCEv2 IPv4, потом RoCEv2 IPv6, потом RoCEv1 IPv4, потом
+RoCEv1 IPv6, потом IB. Авто-выбор GID не поддерживается со старыми версиями
+libibverbs < v32.
+
+Правильный rdma_gid_index для RoCEv2, как правило, 1 (IPv6) или 3 (IPv4).

 ## rdma_mtu

@@ -55,7 +55,7 @@ Examples:
 OSD placement tree is set in a separate etcd key `/vitastor/config/node_placement`
 in the following JSON format:

-`
+```
 {
  "<node name or OSD number>": {
    "level": "<level>",
@@ -63,7 +63,7 @@ in the following JSON format:
  },
  ...
 }
-`
+```

 Here, if a node name is a number then it is assumed to refer to an OSD.
 Level of the OSD is always "osd" and cannot be overriden. You may only
@@ -54,7 +54,7 @@
 Дерево размещения OSD задаётся в отдельном ключе etcd `/vitastor/config/node_placement`
 в следующем JSON-формате:

-`
+```
 {
  "<имя узла или номер OSD>": {
    "level": "<уровень>",
@@ -62,7 +62,7 @@
  },
  ...
 }
-`
+```

 Здесь, если название узла - число, считается, что это OSD. Уровень OSD
 всегда равен "osd" и не может быть переопределён. Для OSD вы можете только
@@ -97,9 +97,9 @@
    it (they have internal SSD cache even though it's not stated in datasheets).

    Setting this parameter to "all" or "small" in OSD parameters requires enabling
-    [disable_journal_fsync](layout-osd.en.yml#disable_journal_fsync) and
-    [disable_meta_fsync](layout-osd.en.yml#disable_meta_fsync), setting it to
-    "all" also requires enabling [disable_data_fsync](layout-osd.en.yml#disable_data_fsync).
+    [disable_journal_fsync](layout-osd.en.md#disable_journal_fsync) and
+    [disable_meta_fsync](layout-osd.en.md#disable_meta_fsync), setting it to
+    "all" also requires enabling [disable_data_fsync](layout-osd.en.md#disable_data_fsync).
    vitastor-disk tried to do that by default, first checking/disabling drive cache.
    If it can't disable drive cache, OSD get initialized with "none".
  info_ru: |
@@ -156,6 +156,6 @@
    указано в спецификациях).

    Указание "all" или "small" в настройках / командной строке OSD требует
-    включения [disable_journal_fsync](layout-osd.ru.yml#disable_journal_fsync) и
-    [disable_meta_fsync](layout-osd.ru.yml#disable_meta_fsync), значение "all"
-    также требует включения [disable_data_fsync](layout-osd.ru.yml#disable_data_fsync).
+    включения [disable_journal_fsync](layout-osd.ru.md#disable_journal_fsync) и
+    [disable_meta_fsync](layout-osd.ru.md#disable_meta_fsync), значение "all"
+    также требует включения [disable_data_fsync](layout-osd.ru.md#disable_data_fsync).
@@ -110,20 +110,22 @@
  type: bool
  default: false
  info: |
-    Do not issue fsyncs to the data device, i.e. do not flush its cache.
-    Safe ONLY if your data device has write-through cache. If you disable
-    the cache yourself using `hdparm` or `scsi_disk/cache_type` then make sure
-    that the cache disable command is run every time before starting Vitastor
-    OSD, for example, in the systemd unit. See also `immediate_commit` option
-    for the instructions to disable cache and how to benefit from it.
+    Do not issue fsyncs to the data device, i.e. do not force it to flush cache.
+    Safe ONLY if your data device has write-through cache or if write-back
+    cache is disabled. If you disable drive cache manually with `hdparm` or
+    writing to `/sys/.../scsi_disk/cache_type` then make sure that you do it
+    every time before starting Vitastor OSD (vitastor-disk does it automatically).
+    See also [immediate_commit](layout-cluster.en.md#immediate_commit)
+    for information about how to benefit from disabled cache.
  info_ru: |
-    Не отправлять fsync-и устройству данных, т.е. не сбрасывать его кэш.
+    Не отправлять fsync-и устройству данных, т.е. не заставлять его сбрасывать кэш.
    Безопасно, ТОЛЬКО если ваше устройство данных имеет кэш со сквозной
-    записью (write-through). Если вы отключаете кэш через `hdparm` или
-    `scsi_disk/cache_type`, то удостоверьтесь, что команда отключения кэша
-    выполняется перед каждым запуском Vitastor OSD, например, в systemd unit-е.
-    Смотрите также опцию `immediate_commit` для инструкций по отключению кэша
-    и о том, как из этого извлечь выгоду.
+    записью (write-through) или если кэш с отложенной записью (write-back) отключён.
+    Если вы отключаете кэш вручную через `hdparm` или запись в `/sys/.../scsi_disk/cache_type`,
+    то удостоверьтесь, что вы делаете это каждый раз перед запуском Vitastor OSD
+    (vitastor-disk делает это автоматически). Смотрите также опцию
+    [immediate_commit](layout-cluster.ru.md#immediate_commit) для информации о том,
+    как извлечь выгоду из отключённого кэша.
 - name: disable_meta_fsync
  type: bool
  default: false
@@ -179,8 +181,7 @@

    Because of this it can actually be beneficial to use SSDs which work well
    with 512 byte sectors and use 512 byte disk_alignment, journal_block_size
-    and meta_block_size. But the only SSD that may fit into this category is
-    Intel Optane (probably, not tested yet).
+    and meta_block_size. But at the moment, no such SSDs are known...

    Clients don't need to be aware of disk_alignment, so it's not required to
    put a modified value into etcd key /vitastor/config/global.
@@ -198,9 +199,8 @@

    Поэтому, на самом деле, может быть выгодно найти SSD, хорошо работающие с
    меньшими, 512-байтными, блоками и использовать 512-байтные disk_alignment,
-    journal_block_size и meta_block_size. Однако единственные SSD, которые
-    теоретически могут попасть в эту категорию - это Intel Optane (но и это
-    пока не проверялось автором).
+    journal_block_size и meta_block_size. Однако на данный момент такие SSD
+    не известны...

    Клиентам не обязательно знать про disk_alignment, так что помещать значение
    этого параметра в etcd в /vitastor/config/global не нужно.
@@ -172,3 +172,27 @@
  info_ru: |
    Использовать старый генератор комбинаций PG, не поддерживающий [level_placement](pool.ru.md#level_placement)
    и [raw_placement](pool.ru.md#raw_placement) для пулов, которые не используют данные функции.
+- name: osd_backfillfull_ratio
+  type: float
+  default: 0.99
+  info: |
+    Monitors try to prevent OSDs becoming 100% full during rebalance or recovery by
+    calculating how much space will be occupied on every OSD after all rebalance
+    and recovery operations finish, and pausing rebalance and recovery if that
+    amount of space exceeds OSD capacity multiplied by the value of this
+    configuration parameter.
+
+    Future used space is calculated by summing space used by all user data blocks
+    (objects) in all PGs placed on a specific OSD, even if some of these objects
+    currently reside on a different set of OSDs.
+  info_ru: |
+    Мониторы стараются предотвратить 100% заполнение OSD в процессе ребаланса
+    или восстановления, рассчитывая, сколько места будет занято на каждом OSD после
+    завершения всех операций ребаланса и восстановления, и приостанавливая
+    ребаланс и восстановление, если рассчитанный объём превышает ёмкость OSD,
+    умноженную на значение данного параметра.
+
+    Будущее занятое место рассчитывается сложением места, занятого всеми
+    пользовательскими блоками данных (объектами) во всех PG, расположенных
+    на конкретном OSD, даже если часть этих объектов в данный момент находится
+    на другом наборе OSD.
@@ -48,11 +48,17 @@
  type: string
  info: |
    RDMA device name to use for Vitastor OSD communications (for example,
-    "rocep5s0f0"). Now Vitastor supports all adapters, even ones without
-    ODP support, like Mellanox ConnectX-3 and non-Mellanox cards.
+    "rocep5s0f0"). If not specified, Vitastor will try to find an RoCE
+    device matching [osd_network](osd.en.md#osd_network), preferring RoCEv2,
+    or choose the first available RDMA device if no RoCE devices are
+    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.

-    Versions up to Vitastor 1.2.0 required ODP which is only present in
-    Mellanox ConnectX >= 4. See also [rdma_odp](#rdma_odp).
+    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.
@@ -64,12 +70,17 @@
    PFC (Priority Flow Control) and ECN (Explicit Congestion Notification).
  info_ru: |
    Название RDMA-устройства для связи с Vitastor OSD (например, "rocep5s0f0").
-    Сейчас 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.
+    Если не указано, Vitastor попробует найти RoCE-устройство, соответствующее
+    [osd_network](osd.en.md#osd_network), предпочитая RoCEv2, или выбрать первое
+    попавшееся RDMA-устройство, если RoCE-устройств нет или если сеть `osd_network`
+    не задана. Также автовыбор не поддерживается со старыми версиями библиотеки
+    libibverbs < v32, например в Debian 10 Buster или CentOS 7.

-    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).
+    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.
@@ -94,23 +105,29 @@
    `ibv_devinfo -v`.
 - name: rdma_gid_index
  type: int
-  default: 0
  info: |
    Global address identifier index of the RDMA device to use. Different GID
    indexes may correspond to different protocols like RoCEv1, RoCEv2 and iWARP.
    Search for "GID" in `ibv_devinfo -v` output to determine which GID index
    you need.

-    **IMPORTANT:** If you want to use RoCEv2 (as recommended) then the correct
-    rdma_gid_index is usually 1 (IPv6) or 3 (IPv4).
+    If not specified, Vitastor will try to auto-select a RoCEv2 IPv4 GID, then
+    RoCEv2 IPv6 GID, then RoCEv1 IPv4 GID, then RoCEv1 IPv6 GID, then IB GID.
+    GID auto-selection is unsupported with libibverbs < v32.
+
+    A correct rdma_gid_index for RoCEv2 is usually 1 (IPv6) or 3 (IPv4).
  info_ru: |
    Номер глобального идентификатора адреса RDMA-устройства, который следует
    использовать. Разным gid_index могут соответствовать разные протоколы связи:
    RoCEv1, RoCEv2, iWARP. Чтобы понять, какой нужен вам - смотрите строчки со
    словом "GID" в выводе команды `ibv_devinfo -v`.

-    **ВАЖНО:** Если вы хотите использовать RoCEv2 (как мы и рекомендуем), то
-    правильный rdma_gid_index, как правило, 1 (IPv6) или 3 (IPv4).
+    Если не указан, Vitastor попробует автоматически выбрать сначала GID,
+    соответствующий RoCEv2 IPv4, потом RoCEv2 IPv6, потом RoCEv1 IPv4, потом
+    RoCEv1 IPv6, потом IB. Авто-выбор GID не поддерживается со старыми версиями
+    libibverbs < v32.
+
+    Правильный rdma_gid_index для RoCEv2, как правило, 1 (IPv6) или 3 (IPv4).
 - name: rdma_mtu
  type: int
  default: 4096
@@ -6,9 +6,18 @@

 # Kubernetes CSI

-Vitastor has a CSI plugin for Kubernetes which supports RWO (and block RWX) volumes.
+Vitastor has a CSI plugin for Kubernetes which supports block-based and VitastorFS-based volumes.

-To deploy it, take manifests from [csi/deploy/](../../csi/deploy/) directory, put your
+Block-based volumes may be formatted and mounted with a normal FS (ext4 or xfs). Such volumes
+only support RWO (ReadWriteOnce) mode.
+
+Block-based volumes may also be left without FS and attached into the container as a block
+device. Such volumes also support RWX (ReadWriteMany) mode.
+
+VitastorFS-based volumes use a clustered file system and support FS-based RWX (ReadWriteMany)
+mode. However, such volumes don't support quotas and snapshots.
+
+To deploy the CSI plugin, take manifests from [csi/deploy/](../../csi/deploy/) directory, put your
 Vitastor configuration in [001-csi-config-map.yaml](../../csi/deploy/001-csi-config-map.yaml),
 configure storage class in [009-storage-class.yaml](../../csi/deploy/009-storage-class.yaml)
 and apply all `NNN-*.yaml` manifests to your Kubernetes installation:
@@ -23,16 +32,16 @@ After that you'll be able to create PersistentVolumes.
 kernel modules enabled (vdpa, vduse, virtio-vdpa). If your distribution doesn't
 have them pre-built - build them yourself ([instructions](../usage/qemu.en.md#vduse)),
 I promise it's worth it :-). When VDUSE is unavailable, CSI driver uses [NBD](../usage/nbd.en.md)
-to map Vitastor devices. NBD is slower and prone to timeout issues: if Vitastor
-cluster becomes unresponsible for more than [nbd_timeout](../config/client.en.md#nbd_timeout),
-the NBD device detaches and breaks pods using it.
+to map Vitastor devices. NBD is slower and, with kernels older than 5.19, unmountable
+if the cluster becomes unresponsible.

 ## Features

 Vitastor CSI supports:
 - Kubernetes starting with 1.20 (or 1.17 for older vitastor-csi <= 1.1.0)
- Filesystem RWO (ReadWriteOnce) volumes. Example: [PVC](../../csi/deploy/example-pvc.yaml), [pod](../../csi/deploy/example-test-pod.yaml)
+- Block-based FS-formatted RWO (ReadWriteOnce) volumes. Example: [PVC](../../csi/deploy/example-pvc.yaml), [pod](../../csi/deploy/example-test-pod.yaml)
 - Raw block RWX (ReadWriteMany) volumes. Example: [PVC](../../csi/deploy/example-pvc-block.yaml), [pod](../../csi/deploy/example-test-pod-block.yaml)
+- VitastorFS-based volumes RWX (ReadWriteMany) volumes. Example: [storage class](../../csi/deploy/example-storage-class-fs.yaml)
 - Volume expansion
 - Volume snapshots. Example: [snapshot class](../../csi/deploy/example-snapshot-class.yaml), [snapshot](../../csi/deploy/example-snapshot.yaml), [clone](../../csi/deploy/example-snapshot-clone.yaml)
 - [VDUSE](../usage/qemu.en.md#vduse) (preferred) and [NBD](../usage/nbd.en.md) device mapping methods
@@ -6,7 +6,17 @@

 # Kubernetes CSI

-У Vitastor есть CSI-плагин для Kubernetes, поддерживающий RWO, а также блочные RWX, тома.
+У Vitastor есть CSI-плагин для Kubernetes, поддерживающий блочные тома и тома на основе
+кластерной ФС VitastorFS.
+
+Блочные тома могут быть отформатированы и примонтированы со стандартной ФС (ext4 или xfs).
+Такие тома поддерживают только режим RWO (ReadWriteOnce, одновременный доступ с одного узла).
+
+Блочные тома также могут не форматироваться и подключаться в контейнер в виде блочного устройства.
+В таком случае их можно подключать в режиме RWX (ReadWriteMany, одновременный доступ с многих узлов).
+
+Тома на основе VitastorFS используют кластерную ФС и поэтому также поддерживают режим RWX
+(ReadWriteMany). Однако, такие тома не поддерживают ограничение размера и снимки.

 Для установки возьмите манифесты из директории [csi/deploy/](../../csi/deploy/), поместите
 вашу конфигурацию подключения к Vitastor в [csi/deploy/001-csi-config-map.yaml](../../csi/deploy/001-csi-config-map.yaml),
@@ -33,6 +43,7 @@ CSI-плагин Vitastor поддерживает:
 - Версии Kubernetes, начиная с 1.20 (или с 1.17 для более старых vitastor-csi <= 1.1.0)
 - Файловые RWO (ReadWriteOnce) тома. Пример: [PVC](../../csi/deploy/example-pvc.yaml), [под](../../csi/deploy/example-test-pod.yaml)
 - Сырые блочные RWX (ReadWriteMany) тома. Пример: [PVC](../../csi/deploy/example-pvc-block.yaml), [под](../../csi/deploy/example-test-pod-block.yaml)
+- Основанные на VitastorFS RWX (ReadWriteMany) тома. Пример: [класс хранения](../../csi/deploy/example-storage-class-fs.yaml)
 - Расширение размера томов
 - Снимки томов. Пример: [класс снимков](../../csi/deploy/example-snapshot-class.yaml), [снимок](../../csi/deploy/example-snapshot.yaml), [клон снимка](../../csi/deploy/example-snapshot-clone.yaml)
 - Способы подключения устройств [VDUSE](../usage/qemu.ru.md#vduse) (предпочитаемый) и [NBD](../usage/nbd.ru.md)
@@ -4,6 +4,8 @@

 [Читать на русском](opennebula.ru.md)

+# OpenNebula
+
 ## Automatic Installation

 OpenNebula plugin is packaged as `vitastor-opennebula` Debian and RPM package since Vitastor 1.9.0. So:
@@ -4,6 +4,8 @@

 [Read in English](opennebula.en.md)

+# OpenNebula
+
 ## Автоматическая установка

 Плагин OpenNebula Vitastor распространяется как Debian и RPM пакет `vitastor-opennebula`, начиная с версии Vitastor 1.9.0. Так что:
@@ -6,19 +6,151 @@

 # Architecture

+- [Server-side components](#server-side-components)
 - [Basic concepts](#basic-concepts)
+- [Client-side components](#client-side-components)
+- [Additional utilities](#additional-utilities)
+- [Overall read/write process](#overall-read-write-process)
+  - [Nuances of request handling](#nuances-of-request-handling)
 - [Similarities to Ceph](#similarities-to-ceph)
 - [Differences from Ceph](#differences-from-ceph)
 - [Implementation Principles](#implementation-principles)

+## Server-side components
+
+- **OSD** (Object Storage Daemon) is a process that directly works with the disk, stores data
+  and serves read/write requests. One OSD serves one disk (or one partition). OSDs talk to etcd
+  and to each other — they receive cluster state from etcd, and send read/write requests for
+  secondary copies of data to other OSDs.
+- **etcd** — clustered key/value database, used as a reliable storage for configuration
+  and high-level cluster state. Etcd is the component that prevents splitbrain in the cluster.
+  Data blocks are not stored in etcd, etcd doesn't participate in data write or read path.
+- **Монитор** — a separate node.js based daemon which monitors the cluster, calculates
+  required configuration changes and saves them to etcd, thus commanding OSDs to apply these
+  changes. Monitor also aggregates cluster statistics. OSD don't talk to monitor, monitor
+  only sends and receives data from etcd.
+
 ## Basic concepts

- OSD (Object Storage Daemon) is a process that stores data and serves read/write requests.
- PG (Placement Group) is a "shard" of the cluster, group of data stored on one set of replicas.
- Pool is a container for data that has equal redundancy scheme and placement rules.
- Monitor is a separate daemon that watches cluster state and handles failures.
- Failure Domain is a group of OSDs that you allow to fail. It's "host" by default.
- Placement Tree groups OSDs in a hierarchy to later split them into Failure Domains.
+- **Pool** is a container for data that has equal redundancy scheme and disk placement rules.
+- **PG (Placement Group)** is a "shard" of the cluster, subdivision unit that has its own
+  set of OSDs for data storage.
+- **Failure Domain** is a group of OSDs, from the simultaneous failure of which you are
+  protected by Vitastor. Default failure domain is "host" (server), but you choose a
+  larger (for example, a rack of servers) or smaller (a single drive) failure domain
+  for every pool.
+- **Placement Tree** (similar to Ceph CRUSH Tree) groups OSDs in a hierarchy to later
+  split them into Failure Domains.
+
+## Client-side components
+
+- **Client library** encapsulates client I/O logic. Client library connects to etcd and to all OSDs,
+  receives cluster state from etcd, sends read and write requests directly to all OSDs. Due
+  to the symmetric distributed architecture, all data blocks (each 128 KB by default) are placed
+  to different OSDs, but clients always know where each data block is stored and connect directly
+  to the right OSD.
+
+All other client-side components are based on the client library:
+
+- **[vitastor-cli](../usage/cli.en.md)** — command-line utility for cluster management.
+  Allows to view cluster state, manage pools and images, i.e. create, modify and remove
+  virtual disks, their snapshots and clones.
+- **[QEMU driver](../usage/qemu.en.md)** — pluggable QEMU module allowing QEMU/KVM virtual
+  machines work with virtual Vitastor disks directly from userspace through the client library,
+  without the need to attach disks as kernel block devices. However, if you want to attach
+  disks, you can also do that with the same driver and [VDUSE](../usage/qemu.en.md#vduse).
+- **[vitastor-nbd](../usage/nbd.en.md)** — utility that allows to attach Vitastor disks as
+  kernel block devices using NBD (Network Block Device), which works more like "BUSE"
+  (Block Device In Userspace). Vitastor doesn't have Linux kernel modules for the same task
+  (at least by now). NBD is an older, non-recommended way to attach disks — you should use
+  VDUSE whenever you can.
+- **[CSI driver](../installation/kubernetes.en.md)** — driver for attaching Vitastor images
+  and VitastorFS subdirectories as Kubernetes persistent volumes. Block-based CSI uses
+  VDUSE (when available) or NBD — images are attached as kernel block devices and mounted
+  into containers. FS-based CSI uses **[vitastor-nfs](../usage/nfs.en.md)**.
+- **Drivers for Proxmox, OpenStack and so on** — pluggable modules for corresponding systems,
+  allowing to use Vitastor as storage in them.
+- **[vitastor-nfs](../usage/nfs.en.md)** — NFS 3.0 server allowing export of two file system variants:
+  the first is a simplified pseudo-FS for file-based access to Vitastor block images (for non-QEMU
+  hypervisors with NFS support), the second is **VitastorFS**, full-featured clustered POSIX FS.
+  Both variants support parallel access from multiple vitastor-nfs servers. In fact, you are
+  not required to setup separate NFS servers at all and use vitastor-nfs mount command on every
+  client node — it starts the NFS server and mounts the FS locally.
+- **[fio driver](../usage/fio.en.md)** — pluggable module for fio disk benchmarking tool for
+  running performance tests on your Vitastor cluster.
+- **vitastor-kv** — client for a key-value DB working over shared block volumes (usual
+  vitastor images). VitastorFS metadata is stored in vitastor-kv.
+
+## Additional utilities
+
+- **vitastor-disk** — a Vitastor OSD disk management tool. You can create, remove,
+  resize and move OSD partitions with it.
+
+## Overall read/write process
+
+- Vitastor stores virtual disks, also named "images" or "inodes".
+- Each image is stored in some pool. Pool specifies storage parameters such as redundancy
+  scheme (replication or EC — erasure codes, i.e. error correction codes), failure domain
+  and restrictions on OSD selection for image data placement. See [Pool configuration](../config/pool.en.md) for details.
+- Each image is split into objects/blocks of fixed size, equal to [block_size](../config/layout-cluster.en.md#block_size)
+  (128 KB by default), multiplied by data part count for EC or 1 for replicas. That is,
+  if a pool uses EC 4+2 coding scheme (4 data parts + 2 parity parts), then, with the
+  default block_size, images are split into 512 KB objects.
+- Client read/write requests are split into parts at object boundaries.
+- Each object is mapped to a PG number it belongs to, by simply taking a remainder of
+  division of its offset by PG count of the image's pool.
+- Client reads primary OSD for all PGs from etcd. Primary OSD for each PG is assigned
+  by the monitor during cluster operation, along with the full PG OSD set.
+- If not already connected, client connects to primary OSDs of all PGs involved in a
+  read/write request and sends parts of the request to them.
+- If a primary OSD is unavailable, client retries connection attempts indefinitely
+  either until it becomes available or until the monitor assigns another OSD as primary
+  for that PG.
+- Client also retries requests if the primary OSD replies with error code EPIPE, meaning
+  that the PG is inactive at this OSD at the moment - for example, when the primary OSD
+  is switched, or if the primary OSD itself loses connection to replicas during request
+  handling.
+- Primary OSD determines where the parts of the object are stored. By default, all objects
+  are assumed to be stored at the target OSD set of a PG, but some of them may be present
+  at a different OSD set if they are degraded or moved, or if the data rebalancing process
+  is active. OSDs doesn't do any network requests, if calculates locations of all objects
+  during PG activation and stores it in memory.
+- Primary OSD handles the request locally when it can - for example, when it's a read
+  from a replicated pool or when it's a read from a EC pool involving only one data part
+  stored on the OSD's local disk.
+- When a request requires reads or writes to additional OSDs, primary OSD uses already
+  established connections to secondary OSDs of the PG to execute these requests. This happens
+  in parallel to local disk operations. All such connections are guaranteed to be already
+  established when the PG is active, and if any of them is dropped, PG is restarted and
+  all current read/write operations to it fail with EPIPE error and are retried by clients.
+- After completing all secondary read/write requests, primary OSD sends the response to
+  the client.
+
+### Nuances of request handling
+
+- If a pool uses erasure codes and some of the OSDs are unavailable, primary OSDs recover
+  data from the remaining parts during read.
+- Each object has a version number. During write, primary OSD first determines the current
+  version of the object. As primary OSD usually stores the object or its part itself, most
+  of the time version is read from the memory of the OSD itself. However, if primary OSD
+  doesn't contain parts of the object, it requests the version number from a secondary OSD
+  which has that part. Such request still doesn't involve reading from the disk though,
+  because object metadata, including version number, is always stored in OSD memory.
+- If a pool uses erasure codes, partial writes of an object require reading other parts of
+  it from secondary OSDs or from the local disk of the primary OSD itself. This is called
+  "read-modify-write" process.
+- If a pool uses erasure codes, two-phase write process is used to get rid of the Write Hole
+  problem: first a new version of object parts is written to all secondary OSDs without
+  removing the previous version, and then, after receiving successful write confirmations
+  from all OSDs, new version is committed and the old one is allowed to be removed.
+- In a pool doesn't use immediate_commit mode, then write requests sent by clients aren't
+  treated as committed to physical media instantly. Clients have to send separate type of
+  requests (SYNC) to commit changes, and before it isn't sent, new versions of data are
+  allowed to be lost if some OSDs die. Thus, when immediate_commit is disabled, clients
+  store copies of all write requests in memory and repeat them from there when the
+  connection to primary OSD is lost. This in-memory copy is removed after a successful
+  SYNC, and to prevent excessive memory usage, clients also do an automatic SYNC
+  every [client_dirty_limit](../config/network.en.md#client_dirty_limit) written bytes.

 ## Similarities to Ceph

@@ -87,5 +219,5 @@
 - Deleting images in a degraded cluster may currently lead to objects reappearing
  after dead OSDs come back, and in case of erasure-coded pools, they may even
  reappear as incomplete. Just repeat the removal request again in this case.
-  This problem will be fixed in the nearest future, the fix is already implemented
-  in the "epoch-deletions" branch.
+  This problem will be fixed in the future, along with the metadata disk storage
+  format update.
@@ -11,6 +11,7 @@
 - [Серверные компоненты](#серверные-компоненты)
 - [Базовые понятия](#базовые-понятия)
 - [Клиентские компоненты](#клиентские-компоненты)
+- [Дополнительные утилиты](#дополнительные-утилиты)
 - [Общий процесс записи и чтения](#общий-процесс-записи-и-чтения)
  - [Особенности обработки запросов](#особенности-обработки-запросов)
 - [Схожесть с Ceph](#схожесть-с-ceph)
@@ -23,8 +24,8 @@
  Один OSD управляет одним диском (или разделом). OSD общаются с etcd и друг с другом — от etcd они
  получают состояние кластера, а друг другу передают запросы записи и чтения вторичных копий данных.
 - **etcd** — кластерная key/value база данных, используется для хранения настроек и верхнеуровневого
-  состояния кластера, а также предотвращения разделения сознания. Блоки данных в etcd не хранятся,
-  в обработке клиентских запросов чтения и записи etcd не участвует.
+  состояния кластера, а также предотвращения разделения сознания (splitbrain). Блоки данных в etcd не
+  хранятся, в обработке клиентских запросов чтения и записи etcd не участвует.
 - **Монитор** — отдельный демон на node.js, рассчитывающий необходимые изменения в конфигурацию
  кластера, сохраняющий эту информацию в etcd и таким образом командующий OSD применить эти изменения.
  Также агрегирует статистику. Контактирует только с etcd, OSD с монитором не общаются.
@@ -34,40 +35,56 @@
 - **Пул (Pool)** — контейнер для данных, имеющих одну и ту же схему избыточности и правила распределения по OSD.
 - **PG (Placement Group)** — "шард", единица деления пулов в кластере, которой назначается свой набор
  OSD для хранения данных (копий или частей объектов).
- **Домен отказа (Failure Domain)** — группа OSD, одновременное падение которых рассматривается
-  как вероятное. По умолчанию это "host" (сервер).
+- **Домен отказа (Failure Domain)** — группа OSD, от одновременного падения которых должен защищать
+  Vitastor. По умолчанию домен отказа — "host" (сервер), но вы можете установить для пула как больший
+  домен отказа (например, стойку серверов), так и меньший (например, отдельный диск).
 - **Дерево распределения** (Placement Tree, в Ceph CRUSH Tree) — иерархическая группировка OSD
  в узлы, которые далее можно использовать как домены отказа.

 ## Клиентские компоненты

- **Клиентская библиотека** — инкапсулирует логику на стороне клиента. Соединяются с etcd и со всеми OSD,
-  от etcd получают состояние кластера, команды чтения и записи отправляют на все OSD напрямую.
+- **Клиентская библиотека** — инкапсулирует логику на стороне клиента. Соединяется с etcd и со всеми OSD,
+  от etcd получает состояние кластера, команды чтения и записи отправляет на все OSD напрямую.
  В силу архитектуры все отдельные блоки данных (по умолчанию по 128 КБ) располагается на разных
  OSD, но клиент устроен так, что всегда точно знает, к какому OSD обращаться, и подключается
  к нему напрямую.

 На базе клиентской библиотеки реализованы все остальные клиенты:

- **vitastor-cli** — утилита командной строки для управления кластером. В данный момент позволяет
-  просматривать общее состояние кластера и управлять образами — т.е. создавать, менять и удалять
-  виртуальные диски, их снимки и клоны.
- **Драйвер QEMU** — подключаемый модуль QEMU, позволяющий QEMU/KVM виртуальным машинам работать
-  с виртуальными дисками Vitastor напрямую из пространства пользователя с помощью клиентской
-  библиотеки, без необходимости отображения дисков в виде блочных устройств. Тот же драйвер
-  позволяет подключать диски в систему через [VDUSE](../usage/qemu.ru.md#vduse).
- **vitastor-nbd** — утилита, позволяющая монтировать образы Vitastor в виде блочных устройств
-  с помощью NBD (Network Block Device), на самом деле скорее работающего как "BUSE"
-  (Block Device In Userspace). Модуля ядра Linux для выполнения той же задачи в Vitastor нет
-  (по крайней мере, пока).
- **CSI драйвер** — драйвер для подключения Vitastor-образов в виде персистентных томов (PV) Kubernetes.
-  Работает через vitastor-nbd — образы отражаются в виде блочных устройств и монтируются
-  в контейнеры.
+- **[vitastor-cli](../usage/cli.ru.md)** — утилита командной строки для управления кластером.
+  Позволяет просматривать общее состояние кластера, управлять пулами и образами — то есть
+  создавать, менять и удалять виртуальные диски, их снимки и клоны.
+- **[Драйвер QEMU](../usage/qemu.ru.md)** — подключаемый модуль QEMU, позволяющий QEMU/KVM
+  виртуальным машинам работать с виртуальными дисками Vitastor напрямую из пространства пользователя
+  с помощью клиентской библиотеки, без необходимости подключения дисков в виде блочных устройств
+  Linux. Если, однако, вы хотите подключать диски в виде блочных устройств, то вы тоже можете
+  сделать это с помощью того же самого драйвера и [VDUSE](../usage/qemu.ru.md#vduse).
+- **[vitastor-nbd](../usage/nbd.ru.md)** — утилита, позволяющая монтировать образы Vitastor
+  в виде блочных устройств с помощью NBD (Network Block Device), на самом деле скорее работающего
+  как "BUSE" (Block Device In Userspace). Модуля ядра Linux для выполнения той же задачи в
+  Vitastor нет (по крайней мере, пока). NBD — более старый и нерекомендуемый способ подключения
+  дисков — вам следует использовать VDUSE всегда, когда это возможно.
+- **[CSI драйвер](../installation/kubernetes.ru.md)** — драйвер для подключения Vitastor-образов
+  и поддиректорий VitastorFS в виде персистентных томов (PV) Kubernetes. Блочный CSI работает через
+  VDUSE (когда это возможно) или через NBD — образы отражаются в виде блочных устройств и монтируются
+  в контейнеры. Файловый CSI использует **[vitastor-nfs](../usage/nfs.ru.md)**.
 - **Драйвера Proxmox, OpenStack и т.п.** — подключаемые модули для соответствующих систем,
  позволяющие использовать Vitastor как хранилище в оных.
- **vitastor-nfs** — утилита, предоставляющая файловый доступ к образам в кластере Vitastor
-  по протоколу NFS 3.0. Предназначена для гипервизоров, не основанных на QEMU и Linux, но при
-  этом поддерживающих NFS.
+- **[vitastor-nfs](../usage/nfs.ru.md)** — NFS 3.0 сервер, предоставляющий два варианта файловой системы:
+  первая — упрощённая для файлового доступа к блочным образам (для не-QEMU гипервизоров, поддерживающих NFS),
+  вторая — VitastorFS, полноценная кластерная POSIX ФС. Оба варианта поддерживают параллельный
+  доступ с нескольких vitastor-nfs серверов. На самом деле можно вообще не выделять
+  отдельные NFS-серверы, а вместо этого использовать команду vitastor-nfs mount, запускающую
+  NFS-сервер прямо на клиентской машине и монтирующую ФС локально.
+- **[Драйвер fio](../usage/fio.ru.md)** — подключаемый модуль для утилиты тестирования
+  производительности дисков fio, позволяющий тестировать Vitastor-кластеры.
+- **vitastor-kv** — клиент для key-value базы данных, работающей поверх разделяемого блочного
+  образа (обычного блочного образа vitastor). Метаданные VitastorFS хранятся именно в vitastor-kv.
+
+## Дополнительные утилиты
+
+- **vitastor-disk** — утилита для разметки дисков под Vitastor OSD. С её помощью можно
+  создавать, удалять, менять размеры или перемещать разделы OSD.

 ## Общий процесс записи и чтения

@@ -98,16 +115,22 @@
  находиться на других OSD, если эти объекты деградированы или перемещены, или идёт процесс
  ребаланса. Запросы для проверки по сети не отправляются, информация о местоположении всех
  объектов рассчитывается первичным OSD при активации PG и хранится в памяти.
- Первичный OSD соединяется (если ещё не соединён) с вторичными OSD, на которых располагаются
-  части объекта, и отправляет им запросы чтения/записи, а также читает/пишет из/в своё локальное
-  хранилище, если сам входит в набор.
+- Когда это возможно, первичный OSD обрабатывает запрос локально. Например, так происходит
+  при чтениях объектов из пулов с репликацией или при чтении из EC пула, затрагивающего
+  только часть, хранимую на диске самого первичного OSD.
+- Когда запрос требует записи или чтения с вторичных OSD, первичный OSD использует заранее
+  установленные соединения с ними для выполнения этих запросов. Это происходит параллельно
+  локальным операциям чтения/записи с диска самого OSD. Так как соединения к вторичным OSD PG
+  устанавливаются при её запуске, то они уже гарантированно установлены, когда PG активна,
+  и если любое из этих соединений отключается, PG перезапускается, а все текущие запросы чтения
+  и записи в неё завершаются с ошибкой EPIPE, после чего повторяются клиентами.
 - После завершения всех вторичных операций чтения/записи первичный OSD отправляет ответ клиенту.

 ### Особенности обработки запросов

 - Если в пуле используются коды коррекции ошибок и при этом часть OSD недоступна, первичный
  OSD при чтении восстанавливает данные из оставшихся частей.
- Каждый объект имеет номер версии. При записи объекта первичный OSD сначала читает из номер
+- Каждый объект имеет номер версии. При записи объекта первичный OSD сначала получает номер
  версии объекта. Так как первичный OSD обычно сам хранит копию или часть объекта, номер
  версии обычно читается из памяти самого OSD. Однако, если ни одна часть обновляемого объекта
  не находится на первичном OSD, для получения номера версии он обращается к одному из вторичных
@@ -115,20 +138,20 @@
  так как метаданные объектов, включая номер версии, все OSD хранят в памяти.
 - Если в пуле используются коды коррекции ошибок, перед частичной записью объекта для вычисления
  чётности зачастую требуется чтение частей объекта с вторичных OSD или с локального диска
-  самого первичного OSD.
- Также, если в пуле используются коды коррекции ошибок, для закрытия Write Hole применяется
+  самого первичного OSD. Это называется процессом "чтение-модификация-запись" (read-modify-write).
+- Если в пуле используются коды коррекции ошибок, для закрытия Write Hole применяется
  двухфазный алгоритм записи: сначала на все вторичные OSD записывается новая версия частей
  объекта, но при этом старая версия не удаляется, а потом, после получения подтверждения
  успешной записи от всех вторичных OSD, новая версия фиксируется и разрешается удаление старой.
- Если в кластере не включён режим immediate_commit, то запросы записи, отправляемые клиентами,
+- Если в пуле не включён режим immediate_commit, то запросы записи, отправляемые клиентами,
  не считаются зафиксированными на физических накопителях сразу. Для фиксации данных клиенты
  должны отдельно отправлять запросы SYNC (отдельный от чтения и записи вид запроса),
  а пока такой запрос не отправлен, считается, что записанные данные могут исчезнуть,
  если соответствующий OSD упадёт. Поэтому, когда режим immediate_commit отключён, все
  запросы записи клиенты копируют в памяти и при потере соединения и повторном соединении
-  с OSD повторяют из памяти. Скопированные в память данные удаляются при успешном fsync,
+  с OSD повторяют из памяти. Скопированные в память данные удаляются при успешном SYNC,
  а чтобы хранение этих данных не приводило к чрезмерному потреблению памяти, клиенты
-  автоматически выполняют fsync каждые [client_dirty_limit](../config/network.ru.md#client_dirty_limit)
+  автоматически выполняют SYNC каждые [client_dirty_limit](../config/network.ru.md#client_dirty_limit)
  записанных байт.

 ## Схожесть с Ceph
@@ -205,5 +228,5 @@
 - Удаление образов в деградированном кластере может в данный момент приводить к повторному
  "появлению" удалённых объектов после поднятия отключённых OSD, причём в случае EC-пулов,
  объекты могут появиться в виде "неполных". Если вы столкнётесь с такой ситуацией, просто
-  повторите запрос удаления. Исправление этой проблемы уже реализовано в ветке "epoch-deletions"
-  и вскоре будет включено в релиз.
+  повторите запрос удаления. Данная проблема будет исправлена в будущем вместе с обновлением
+  дискового формата хранения метаданных.
@@ -36,6 +36,7 @@
 - [Clustered file system](../usage/nfs.en.md#vitastorfs)
 - [Experimental internal etcd replacement - antietcd](../config/monitor.en.md#use_antietcd)
 - [Built-in Prometheus metric exporter](../config/monitor.en.md#enable_prometheus)
+- [NFS RDMA support](../usage/nfs.en.md#rdma) (probably also usable for GPUDirect)

 ## Plugins and tools

@@ -38,6 +38,7 @@
 - [Кластерная файловая система](../usage/nfs.ru.md#vitastorfs)
 - [Экспериментальная встроенная замена etcd - antietcd](../config/monitor.ru.md#use_antietcd)
 - [Встроенный Prometheus-экспортер метрик](../config/monitor.ru.md#enable_prometheus)
+- [Поддержка NFS RDMA](../usage/nfs.ru.md#rdma) (вероятно, также подходящая для GPUDirect)

 ## Драйверы и инструменты

@@ -32,7 +32,7 @@

 - 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 D7-P5500/P5600, Intel Optane, Kingston DC1000B/DC1500M
+  Intel DC-P3700/P4500/P4600, Intel D5-P4320, Intel D7-P5500/P5600, Intel Optane, Kingston DC1000B/DC1500M
 - HDD: HGST Ultrastar, Toshiba MG, Seagate EXOS

 ## Configure monitors
@@ -22,7 +22,7 @@
  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 MG08, Seagate EXOS или что-то похожее. Если такого кэша у ваших дисков нет,
+  Toshiba MG, Seagate EXOS или что-то похожее. Если такого кэша у ваших дисков нет,
  обязательно возьмите SSD под метаданные и журнал (маленькие, буквально 2 ГБ на 1 ТБ HDD-места).
 - Возьмите быструю сеть, минимум 10 гбит/с. Идеал - что-то вроде Mellanox ConnectX-4 с RoCEv2.
 - Для лучшей производительности отключите энергосбережение CPU: `cpupower idle-set -D 0 && cpupower frequency-set -g performance`.
@@ -32,8 +32,8 @@

 - 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 D7-P5500/P5600, Intel Optane, Kingston DC1000B/DC1500M
- HDD: HGST Ultrastar, Toshiba MG06/MG07/MG08, Seagate EXOS
+  Intel DC-P3700/P4500/P4600, Intel D5-P4320, Intel D7-P5500/P5600, Intel Optane, Kingston DC1000B/DC1500M
+- HDD: HGST Ultrastar, Toshiba MG, Seagate EXOS

 ## Настройте мониторы

@@ -58,8 +58,9 @@ and during switching primary OSD of PGs.

 **starting**, **repeering**, **stopping** states normally almost aren't visible at all.
 If you notice them for any noticeable time — chances are some operations on some OSDs hung.
-Search for "slow op" in OSD logs to find them — operations hung for more than
-[slow_log_interval](../config/osd.en.md#slow_log_interval) are logged as "slow ops".
+Check `vitastor-cli status` and search for "slow op" in OSD logs to find them — operations
+hung for more than [slow_log_interval](../config/osd.en.md#slow_log_interval) are logged as
+"slow ops" and displayed in `status`.

 State transition diagram:

@@ -169,6 +170,25 @@ Upgrading is performed without stopping clients (VMs/containers), you just need
 upgrade and restart servers one by one. However, ideally you should restart VMs too
 to make them use the new version of the client library.

+### 1.7.x to 1.8.0
+
+It's recommended to upgrade from version <= 1.7.x to version >= 1.8.0 with full downtime,
+i.e. you should first stop clients and then the cluster (OSDs and monitor), because 1.8.0
+includes a fix for etcd event stream inconsistency which could lead to "incomplete" objects
+appearing in EC pools, and in rare cases, probably, even to data corruption during mass OSD
+restarts. It doesn't mean that you WILL hit this problem if you upgrade without full downtime,
+but it's better to secure yourself against it.
+
+Also, if you upgrade version from <= 1.7.x to version >= 1.8.0, BUT <= 1.9.0: restart all clients
+(VMs and so on), otherwise they will hang when monitor clears old PG configuration key,
+which happens 24 hours after upgrade.
+
+This is fixed in 1.9.1. So, after upgrading version <= 1.7.x directly to version >= 1.9.1,
+you DO NOT have to restart all old clients immediately - they will work like before until
+you decide to upgrade them too. The downside is that you'll have to remove the old PG
+configuration key (`/vitastor/config/pgs`) from etcd by hand when you make sure that all
+your clients are restarted.
+
 ### 1.1.x to 1.2.0

 Upgrading version <= 1.1.x to version >= 1.2.0, if you use EC n+k with k>=2, is recommended
@@ -56,9 +56,9 @@ OSD, на протяжении небольшого периода времен
 Состояния **starting**, **repeering**, **stopping** в норме практически не заметны вообще,
 PG должны очень быстро переходить из них в другие. Если эти состояния заметны
 хоть сколько-то значительное время — вероятно, какие-то операции на каких-то OSD зависли.
-Чтобы найти их, ищите "slow op" в журналах OSD — операции, зависшие дольше,
-чем на [slow_log_interval](../config/osd.ru.md#slow_log_interval), записываются в
-журналы OSD как "slow op".
+Чтобы найти их, посморите `vitastor-cli status` и поищите слова "slow op" в журналах OSD —
+операции, зависшие дольше, чем на [slow_log_interval](../config/osd.ru.md#slow_log_interval),
+записываются в журналы OSD как "slow op" и отображаются в `status`.

 Диаграмма переходов:

@@ -166,6 +166,25 @@ done
 достаточно обновлять серверы по одному. Однако, конечно, чтобы запущенные виртуальные машины
 начали использовать новую версию клиентской библиотеки, их тоже нужно перезапустить.

+### 1.7.x -> 1.8.0
+
+Обновляться с версий <= 1.7.x до версий >= 1.8.0 рекомендуется с полной остановкой
+сначала клиентов, а затем кластера, так как в 1.8.0 исправлена проблема (неконсистентность
+потоков событий от etcd), способная приводить к появлению incomplete объектов в EC-пулах
+и, хоть и редко, но даже к повреждению данных при массовых перезапусках OSD. Если вы
+обновляетесь без полной остановки - это не значит, что вы обязательно столкнётесь с этой
+проблемой, но лучше подстраховаться.
+
+Также, если вы обновляетесь с версии <= 1.7.x до версии >= 1.8.0, НО <= 1.9.0: перезапустите всех
+клиентов (процессы виртуальных машин можно перезапустить путём миграции на другой сервер),
+иначе они зависнут, когда монитор удалит старый ключ конфигурации PG, что происходит через
+24 часа после обновления.
+
+Однако, это исправлено в 1.9.1. Так что, если вы обновляетесь с <= 1.7.x сразу до >= 1.9.1,
+вам НЕ нужно сразу перезапускать всех клиентов - они будут работать, как раньше. Минус,
+правда, в том, что старый ключ конфигурации PG (`/vitastor/config/pgs`) будет нужно удалить
+вам из etcd вручную - после того, как вы убедитесь, что все клиенты перезапущены.
+
 ### 1.1.x -> 1.2.0

 Обновляться с версий <= 1.1.x до версий >= 1.2.0, если вы используете EC n+k и k>=2,
@@ -13,6 +13,7 @@ It supports the following commands:
 - [prepare](#prepare)
 - [upgrade-simple](#upgrade-simple)
 - [resize](#resize)
+- [raw-resize](#raw-resize)
 - [start/stop/restart/enable/disable](#start/stop/restart/enable/disable)
 - [purge](#purge)
 - [read-sb](#read-sb)
@@ -50,12 +51,16 @@ Options (automatic mode):
 --osd_per_disk <N>
  Create <N> OSDs on each disk (default 1)
 --hybrid
-  Prepare hybrid (HDD+SSD) OSDs using provided devices. SSDs will be used for
-  journals and metadata, HDDs will be used for data. Partitions for journals and
-  metadata will be created automatically. Whether disks are SSD or HDD is decided
-  by the `/sys/block/.../queue/rotational` flag. In hybrid mode, default object
-  size is 1 MB instead of 128 KB, default journal size is 1 GB instead of 32 MB,
-  and throttle_small_writes is enabled by default.
+  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,
+  but you can override this behaviour with --fast-devices option. Journal and metadata
+  partitions will be created automatically. In the default mode, SSD and HDD disks
+  are distinguished by the `/sys/block/.../queue/rotational` flag. When HDDs are used
+  for data in hybrid mode, default block_size is 1 MB instead of 128 KB, default journal
+  size is 1 GB instead of 32 MB, and throttle_small_writes is enabled by default.
+--fast-devices /dev/nvmeX,/dev/nvmeY
+  In --hybrid mode, use these devices for journal and metadata instead of auto-detecting
+  and extracting them from the main [devices...] list.
 --disable_data_fsync auto
  Disable data device cache and fsync (1/yes/true = on, default auto)
 --disable_meta_fsync auto
@@ -127,25 +132,49 @@ Requires the `sfdisk` utility.

 ## resize

-`vitastor-disk resize <ALL_OSD_PARAMETERS> <NEW_LAYOUT> [--iodepth 32]`
+`vitastor-disk resize <osd_num>|<osd_device> [OPTIONS]`

-Resize data area and/or rewrite/move journal and metadata.
+Resize data area and/or move journal and metadata:
+
+| <!-- -->                  | <!-- -->                               |
+|---------------------------|----------------------------------------|
+| `--move-journal TARGET`   | move journal to `TARGET`               |
+| `--move-meta TARGET`      | move metadata to `TARGET`              |
+| `--journal-size NEW_SIZE` | resize journal to `NEW_SIZE`           |
+| `--data-size NEW_SIZE`    | resize data device to `NEW_SIZE`       |
+| `--dry-run`               | only show new layout, do not apply it  |
+
+`NEW_SIZE` may include k/m/g/t suffixes.
+
+`TARGET` may be one of:
+
+| <!-- -->       | <!-- -->                                                                 |
+|----------------|--------------------------------------------------------------------------|
+| `<partition>`  | move journal/metadata to an existing GPT partition                       |
+| `<raw_device>` | create a GPT partition on `<raw_device>` and move journal/metadata to it |
+| `""`           | (empty string) move journal/metadata back to the data device             |
+
+## raw-resize
+
+`vitastor-disk raw-resize <ALL_OSD_PARAMETERS> <NEW_LAYOUT> [--iodepth 32]`
+
+Resize data area and/or rewrite/move journal and metadata (manual format).

 `ALL_OSD_PARAMETERS` must include all (at least all disk-related)
 parameters from OSD command line (i.e. from systemd unit or superblock).

 `NEW_LAYOUT` may include new disk layout parameters:

-```
--new_data_offset SIZE     resize data area so it starts at SIZE
--new_data_len SIZE        resize data area to SIZE bytes
--new_meta_device PATH     use PATH for new metadata
--new_meta_offset SIZE     make new metadata area start at SIZE
--new_meta_len SIZE        make new metadata area SIZE bytes long
--new_journal_device PATH  use PATH for new journal
--new_journal_offset SIZE  make new journal area start at SIZE
--new_journal_len SIZE     make new journal area SIZE bytes long
-```
+| <!-- -->                    | <!-- -->                                  |
+|-----------------------------|-------------------------------------------|
+| `--new_data_offset SIZE`    | resize data area so it starts at `SIZE`   |
+| `--new_data_len SIZE`       | resize data area to `SIZE` bytes          |
+| `--new_meta_device PATH`    | use `PATH` for new metadata               |
+| `--new_meta_offset SIZE`    | make new metadata area start at `SIZE`    |
+| `--new_meta_len SIZE`       | make new metadata area `SIZE` bytes long  |
+| `--new_journal_device PATH` | use `PATH` for new journal                |
+| `--new_journal_offset SIZE` | make new journal area start at `SIZE`     |
+| `--new_journal_len SIZE`    | make new journal area `SIZE` bytes long   |

 SIZE may include k/m/g/t suffixes. If any of the new layout parameter
 options are not specified, old values will be used.
@@ -217,10 +246,14 @@ Intended for use from startup scripts (i.e. from systemd units).

 ## dump-journal

+`vitastor-disk dump-journal [OPTIONS] <osd_device>`
+
 `vitastor-disk dump-journal [OPTIONS] <journal_file> <journal_block_size> <offset> <size>`

 Dump journal in human-readable or JSON (if `--json` is specified) format.

+You can specify any OSD device (data, metadata or journal), or the layout manually.
+
 Options:

 ```
@@ -233,23 +266,35 @@ Options:

 ## write-journal

+`vitastor-disk write-journal <osd_device>`
+
 `vitastor-disk write-journal <journal_file> <journal_block_size> <bitmap_size> <offset> <size>`

 Write journal from JSON taken from standard input in the same format as produced by
 `dump-journal --json --format data`.

+You can specify any OSD device (data, metadata or journal), or the layout manually.
+
 ## dump-meta

+`vitastor-disk dump-meta <osd_device>`
+
 `vitastor-disk dump-meta <meta_file> <meta_block_size> <offset> <size>`

 Dump metadata in JSON format.

+You can specify any OSD device (data, metadata or journal), or the layout manually.
+
 ## write-meta

+`vitastor-disk write-meta <osd_device>`
+
 `vitastor-disk write-meta <meta_file> <offset> <size>`

 Write metadata from JSON taken from standard input in the same format as produced by `dump-meta`.

+You can specify any OSD device (data, metadata or journal), or the layout manually.
+
 ## simple-offsets

 `vitastor-disk simple-offsets <device>`
@@ -13,6 +13,7 @@ vitastor-disk - инструмент командной строки для уп
 - [prepare](#prepare)
 - [upgrade-simple](#upgrade-simple)
 - [resize](#resize)
+- [raw-resize](#raw-resize)
 - [start/stop/restart/enable/disable](#start/stop/restart/enable/disable)
 - [purge](#purge)
 - [read-sb](#read-sb)
@@ -50,12 +51,17 @@ vitastor-disk - инструмент командной строки для уп
 --osd_per_disk <N>
  Создавать по несколько (<N>) OSD на каждом диске (по умолчанию 1)
 --hybrid
-  Инициализировать гибридные (HDD+SSD) OSD на указанных дисках. SSD будут
-  использованы для журналов и метаданных, а HDD - для данных. Разделы для журналов
-  и метаданных будут созданы автоматически. Является ли диск SSD или HDD, определяется
-  по флагу `/sys/block/.../queue/rotational`. В гибридном режиме по умолчанию
-  используется размер объекта 1 МБ вместо 128 КБ, размер журнала 1 ГБ вместо 32 МБ
-  и включённый throttle_small_writes.
+  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,
+  но вы можете поменять это поведение опцией --fast-devices. Разделы для журналов
+  и метаданных будут созданы автоматически. В режиме по умолчанию SSD и HDD-диски
+  различаются по флагу `/sys/block/.../queue/rotational`. Когда в гибридном режиме
+  для данных используются HDD, по умолчанию размер блока устанавливается 1 МБ вместо
+  128 КБ, размер журнала 1 ГБ вместо 32 МБ, и throttle_small_writes включается по
+  умолчанию.
+--fast-devices /dev/nvmeX,/dev/nvmeY
+  Использовать данные диски для журналов и метаданных в гибридном режиме вместо их
+  автоопределения и извлечения из основного списка [devices...].
 --disable_data_fsync auto
  Отключать кэш и fsync-и для устройств данных. (1/yes/true = да, по умолчанию автоопределение)
 --disable_meta_fsync auto
@@ -129,27 +135,51 @@ throttle_target_mbs, throttle_target_parallelism, throttle_threshold_us.

 ## resize

-`vitastor-disk resize <ALL_OSD_PARAMETERS> <NEW_LAYOUT> [--iodepth 32]`
+`vitastor-disk resize <osd_num>|<osd_device> [OPTIONS]`

-Изменить размер области данных и/или переместить журнал и метаданные.
+Изменить размер области данных и/или переместить журнал и метаданные:

-В `ALL_OSD_PARAMETERS` нужно указать все относящиеся к диску параметры OSD
+| <!-- -->                      | <!-- -->                                       |
+|-------------------------------|------------------------------------------------|
+| `--move-journal ЦЕЛЬ`         | переместить журнал на `ЦЕЛЬ`                   |
+| `--move-meta ЦЕЛЬ`            | переместить метаданные на `ЦЕЛЬ`               |
+| `--journal-size НОВЫЙ_РАЗМЕР` | изменить размер журнала на `НОВЫЙ_РАЗМЕР`      |
+| `--data-size НОВЫЙ_РАЗМЕР`    | изменить размер диска данных на `НОВЫЙ_РАЗМЕР` |
+| `--dry-run`                   | показать новые параметры, но не применять их   |
+
+`НОВЫЙ_РАЗМЕР` может быть указан с суффиксами k/m/g/t (кило/мега/гига/терабайт).
+
+`ЦЕЛЬ` может быть одним из:
+
+| <!-- -->        | <!-- -->                                                                            |
+|-----------------|-------------------------------------------------------------------------------------|
+| `<раздел>`      | переместить журнал/метаданные на существующий GPT-раздел                            |
+| `<полный_диск>` | создать GPT-раздел на диске `<полный_диск>` и переместить журнал/метаданные на него |
+| `""`            | (пустая строка) переместить журнал/метаданные обратно на диск данных                |
+
+## raw-resize
+
+`vitastor-disk raw-resize <ВСЕ_ПАРАМЕТРЫ_OSD> <НОВЫЕ_РАЗМЕРЫ> [--iodepth 32]`
+
+Изменить размер области данных и/или переместить журнал и метаданные (ручной формат).
+
+В `ВСЕ_ПАРАМЕТРЫ_OSD` нужно указать все относящиеся к диску параметры OSD
 из суперблока OSD или из файла сервиса systemd (в старых версиях).

-В `NEW_LAYOUT` нужно указать новые параметры расположения данных:
+В `НОВЫЕ_РАЗМЕРЫ` нужно указать новые параметры расположения данных:

-```
--new_data_offset РАЗМЕР     сдвинуть начало области данных на РАЗМЕР байт
--new_data_len РАЗМЕР        изменить размер области данных до РАЗМЕР байт
--new_meta_device ПУТЬ       использовать ПУТЬ как новое устройство метаданных
--new_meta_offset РАЗМЕР     разместить новые метаданные по смещению РАЗМЕР байт
--new_meta_len РАЗМЕР        сделать новые метаданные размером РАЗМЕР байт
--new_journal_device ПУТЬ    использовать ПУТЬ как новое устройство журнала
--new_journal_offset РАЗМЕР  разместить новый журнал по смещению РАЗМЕР байт
--new_journal_len РАЗМЕР     сделать новый журнал размером РАЗМЕР байт
-```
+| <!-- -->                      | <!-- -->                                              |
+|-------------------------------|-------------------------------------------------------|
+| `--new_data_offset РАЗМЕР`    | сдвинуть начало области данных на `РАЗМЕР` байт       |
+| `--new_data_len РАЗМЕР`       | изменить размер области данных до `РАЗМЕР` байт       |
+| `--new_meta_device ПУТЬ`      | использовать `ПУТЬ` как новое устройство метаданных   |
+| `--new_meta_offset РАЗМЕР`    | разместить новые метаданные по смещению `РАЗМЕР` байт |
+| `--new_meta_len РАЗМЕР`       | сделать новые метаданные размером `РАЗМЕР` байт       |
+| `--new_journal_device ПУТЬ`   | использовать `ПУТЬ` как новое устройство журнала      |
+| `--new_journal_offset РАЗМЕР` | разместить новый журнал по смещению `РАЗМЕР` байт     |
+| `--new_journal_len РАЗМЕР`    | сделать новый журнал размером `РАЗМЕР` байт           |

-РАЗМЕР может быть указан с суффиксами k/m/g/t. Если любой из новых параметров
+`РАЗМЕР` может быть указан с суффиксами k/m/g/t. Если любой из новых параметров
 расположения не указан, он принимается равным старому значению.

 ## start/stop/restart/enable/disable
@@ -224,10 +254,15 @@ OSD отключены fsync-и.

 ## dump-journal

+`vitastor-disk dump-journal <osd_device>`
+
 `vitastor-disk dump-journal [OPTIONS] <journal_file> <journal_block_size> <offset> <size>`

 Вывести журнал в человекочитаемом или в JSON (с опцией `--json`) виде.

+Вы можете указать любой раздел OSD - данных, журнала или метаданных - либо указать все
+параметры расположения вручную.
+
 Опции:

 ```
@@ -240,22 +275,37 @@ OSD отключены fsync-и.

 ## write-journal

+`vitastor-disk write-journal <osd_device>`
+
 `vitastor-disk write-journal <journal_file> <journal_block_size> <bitmap_size> <offset> <size>`

 Записать журнал из JSON со стандартного ввода в формате, аналогичном `dump-journal --json --format data`.

+Вы можете указать любой раздел OSD - данных, журнала или метаданных - либо указать все
+параметры расположения вручную.
+
 ## dump-meta

+`vitastor-disk dump-meta <osd_device>`
+
 `vitastor-disk dump-meta <meta_file> <meta_block_size> <offset> <size>`

 Вывести метаданные в формате JSON.

+Вы можете указать любой раздел OSD - данных, журнала или метаданных - либо указать все
+параметры расположения вручную.
+
 ## write-meta

+`vitastor-disk write-meta <osd_device>`
+
 `vitastor-disk write-meta <meta_file> <offset> <size>`

 Записать метаданные из JSON со стандартного ввода в формате, аналогичном `dump-meta`.

+Вы можете указать любой раздел OSD - данных, журнала или метаданных - либо указать все
+параметры расположения вручную.
+
 ## simple-offsets

 `vitastor-disk simple-offsets <device>`
@@ -96,7 +96,7 @@ Example output (JSON format):
 vitastor-nbd netlink-map [/dev/nbdN] (--image <image> | --pool <pool> --inode <inode> --size <size in bytes>)
 ```

-On recent kernel versions it's also possinle to map NBD devices using netlink interface.
+On recent kernel versions it's also possible to map NBD devices using netlink interface.

 This is an experimental feature because it doesn't solve all issues of NBD. Differences from regular ioctl-based 'map':

@@ -111,6 +111,21 @@ settings, because Vitastor NFS proxy doesn't keep uncommitted data in memory
 with these settings. But it may even work without `immediate_commit=all` because
 the Linux NFS client repeats all uncommitted writes if it loses the connection.

+## RDMA
+
+vitastor-nfs supports NFS over RDMA, which, in theory, should also allow to use
+VitastorFS from GPUDirect.
+
+You can test NFS-RDMA even if you don't have an RDMA NIC using SoftROCE:
+
+1. First, add SoftROCE device on both servers: `rdma link add rxe0 type rxe netdev eth0`.
+   Here, `rdma` utility is a part the iproute2 package, and `eth0` should be replaced with
+   the name of your Ethernet NIC.
+
+2. Start vitastor-nfs with RDMA: `vitastor-nfs start (--fs <NAME> | --block) --pool <POOL> --port 20049 --nfs_rdma 20049 --portmap 0`
+
+3. Mount the FS: `mount 192.168.0.10:/mnt/test/ /mnt/vita/ -o port=20049,mountport=20049,nfsvers=3,soft,nolock,rdma`
+
 ## Commands

 ### mount
@@ -131,11 +146,16 @@ The server will be automatically stopped when the FS is unmounted.

 Start network NFS server. Options:

-| <!-- -->        | <!-- -->                                                   |
-|-----------------|------------------------------------------------------------|
-| `--bind <IP>`   | bind service to \<IP> address (default 0.0.0.0)            |
-| `--port <PORT>` | use port \<PORT> for NFS services (default is 2049)        |
-| `--portmap 0`   | do not listen on port 111 (portmap/rpcbind, requires root) |
+| <!-- -->               | <!-- -->                                                                                                                    |
+|------------------------|-----------------------------------------------------------------------------------------------------------------------------|
+| `--bind <IP>`          | bind service to \<IP> address (default 0.0.0.0)                                                                             |
+| `--port <PORT>`        | use port \<PORT> for NFS services (default is 2049). Specify "auto" to auto-select and print port                           |
+| `--portmap 0`          | do not listen on port 111 (portmap/rpcbind, requires root)                                                                  |
+| `--nfs_rdma <PORT>`    | enable NFS-RDMA at RDMA-CM port \<PORT> (you can try 20049). If RDMA is enabled and --port is set to 0, TCP will be disabled |
+| `--nfs_rdma_credit 16` | maximum operation credit for RDMA clients (max iodepth)                                                                     |
+| `--nfs_rdma_send 1024` | maximum RDMA send operation count (should be larger than iodepth)                                                           |
+| `--nfs_rdma_alloc 1M`  | RDMA memory allocation rounding                                                                                             |
+| `--nfs_rdma_gc 64M`    | maximum unused RDMA buffers                                                                                                 |

 ### upgrade

@@ -156,17 +176,17 @@ behind. Defragmentation removes garbage and moves data still in use to new volum

 Options:

-| <!-- -->                 | <!-- -->                                                                |
-|--------------------------|------------------------------------------------------------------------ |
-| --volume_untouched 86400 | Defragment volumes last appended to at least this number of seconds ago |
-| --defrag_percent 50      | Defragment volumes with at least this % of removed data                 |
-| --defrag_block_count 16  | Read this number of pool blocks at once during defrag                   |
-| --defrag_iodepth 16      | Move up to this number of files in parallel during defrag               |
-| --trace                  | Print verbose defragmentation status                                    |
-| --dry-run                | Skip modifications, only print status                                   |
-| --recalc-stats           | Recalculate all volume statistics                                       |
-| --include-empty          | Include old and empty volumes; make sure to restart NFS servers before using it |
-| --no-rm                  | Move, but do not delete data                                            |
+| <!-- -->                   | <!-- -->                                                                |
+|----------------------------|------------------------------------------------------------------------ |
+| `--volume_untouched 86400` | Defragment volumes last appended to at least this number of seconds ago |
+| `--defrag_percent 50`      | Defragment volumes with at least this % of removed data                 |
+| `--defrag_block_count 16`  | Read this number of pool blocks at once during defrag                   |
+| `--defrag_iodepth 16`      | Move up to this number of files in parallel during defrag               |
+| `--trace`                  | Print verbose defragmentation status                                    |
+| `--dry-run`                | Skip modifications, only print status                                   |
+| `--recalc-stats`           | Recalculate all volume statistics                                       |
+| `--include-empty`          | Include old and empty volumes; make sure to restart NFS servers before using it |
+| `--no-rm`                  | Move, but do not delete data                                            |

 ## Common options

@@ -116,6 +116,21 @@ JSON-формате :-). Для инспекции содержимого БД
 даже без `immediate_commit=all`, потому что NFS-клиент ядра Linux повторяет все
 незафиксированные запросы при потере соединения.

+## RDMA
+
+vitastor-nfs поддерживает NFS через RDMA. В теории это также должно позволять использовать
+VitastorFS из GPUDirect.
+
+Вы можете протестировать NFS-RDMA, даже если у вас нет RDMA-карты, с помощью SoftROCE:
+
+1. Сначала создайте SoftROCE устройства на обоих тестовых серверах: `rdma link add rxe0 type rxe netdev eth0`.
+   Утилита `rdma` входит в состав пакета iproute2, а `eth0` вам нужно заменить на имя своей
+   сетевой карты.
+
+2. Запустите vitastor-nfs с RDMA: `vitastor-nfs start (--fs <NAME> | --block) --pool <POOL> --port 20049 --nfs_rdma 20049 --portmap 0`
+
+3. Смонтируйте ФС: `mount 192.168.0.10:/mnt/test/ /mnt/vita/ -o port=20049,mountport=20049,nfsvers=3,soft,nolock,rdma`
+
 ## Команды

 ### mount
@@ -136,11 +151,16 @@ JSON-формате :-). Для инспекции содержимого БД

 Запустить сетевой NFS-сервер. Опции:

-| <!-- -->        | <!-- -->                                                              |
-|-----------------|-----------------------------------------------------------------------|
-| `--bind <IP>`   | принимать соединения по адресу \<IP> (по умолчанию 0.0.0.0 - на всех) |
-| `--port <PORT>` | использовать порт \<PORT> для NFS-сервисов (по умолчанию 2049)        |
-| `--portmap 0`   | отключить сервис portmap/rpcbind на порту 111 (по умолчанию включён и требует root привилегий) |
+| <!-- -->               | <!-- -->                                                                                                                    |
+|------------------------|-----------------------------------------------------------------------------------------------------------------------------|
+| `--bind <IP>`          | принимать соединения по адресу \<IP> (по умолчанию 0.0.0.0 - на всех)                                                       |
+| `--port <PORT>`        | использовать порт \<PORT> для NFS-сервисов (по умолчанию 2049). Укажите "auto", чтобы выбрать и напечатать случайный порт   |
+| `--portmap 0`          | отключить сервис portmap/rpcbind на порту 111 (по умолчанию включён и требует root привилегий)                              |
+| `--nfs_rdma <PORT>`    | включить NFS-RDMA на порту RDMA-CM \<PORT> (попробуйте 20049). Если RDMA включено и указано `--port 0`, TCP будет отключено |
+| `--nfs_rdma_credit 16` | максимальный "кредит", глубина очереди для NFS-клиентов                                                                     |
+| `--nfs_rdma_send 1024` | максимальное число операций RDMA отправки (должно быть больше nfs_rdma_credit)                                              |
+| `--nfs_rdma_alloc 1M`  | округление выделения памяти для RDMA-клиентов                                                                               |
+| `--nfs_rdma_gc 64M`    | максимальный объём неиспользуемой памяти RDMA-клиентом перед освобождением                                                  |

 ### upgrade

@@ -164,17 +184,17 @@ JSON-формате :-). Для инспекции содержимого БД

 Опции:

-| <!-- -->                 | <!-- -->                                                                |
-|--------------------------|------------------------------------------------------------------------ |
-| --volume_untouched 86400 | Дефрагментировать только тома, в которые уже не писали это число секунд |
-| --defrag_percent 50      | Дефрагментировать только тома, в которых этот % данных удалён           |
-| --defrag_block_count 16  | Читать это количество блоков пула за один раз                           |
-| --defrag_iodepth 16      | Перемещать одновременно до этого числа файлов                           |
-| --trace                  | Печатать детальную статистику дефрагментации                            |
-| --dry-run                | Не производить никаких изменений, только описать выполняемые действия   |
-| --recalc-stats           | Пересчитать и сохранить статистику всех томов                           |
-| --include-empty          | Дефрагментировать старые и пустые тома; обязательно перезапустите NFS-сервера после использования этой опции |
-| --no-rm                  | Перемещать, но не удалять данные                                        |
+| <!-- -->                   | <!-- -->                                                                |
+|----------------------------|------------------------------------------------------------------------ |
+| `--volume_untouched 86400` | Дефрагментировать только тома, в которые уже не писали это число секунд |
+| `--defrag_percent 50`      | Дефрагментировать только тома, в которых этот % данных удалён           |
+| `--defrag_block_count 16`  | Читать это количество блоков пула за один раз                           |
+| `--defrag_iodepth 16`      | Перемещать одновременно до этого числа файлов                           |
+| `--trace`                  | Печатать детальную статистику дефрагментации                            |
+| `--dry-run`                | Не производить никаких изменений, только описать выполняемые действия   |
+| `--recalc-stats`           | Пересчитать и сохранить статистику всех томов                           |
+| `--include-empty`          | Дефрагментировать старые и пустые тома; обязательно перезапустите NFS-сервера после использования этой опции |
+| `--no-rm`                  | Перемещать, но не удалять данные                                        |

 ## Общие опции

@@ -151,9 +151,9 @@ Example performance comparison:
 To try VDUSE you need at least Linux 5.15, built with VDUSE support
 (CONFIG_VDPA=m, CONFIG_VDPA_USER=m, CONFIG_VIRTIO_VDPA=m).

-Debian Linux kernels have these options disabled by now, so if you want to try it on Debian,
-use a kernel from Ubuntu [kernel-ppa/mainline](https://kernel.ubuntu.com/~kernel-ppa/mainline/), Proxmox,
-or build modules for Debian kernel manually:
+Debian Linux kernels had these options disabled until 6.6, so make sure you install a newer kernel
+(from bookworm-backports, trixie or newer Debian version) if you want to try VDUSE. You can also
+build modules for an existing kernel manually:

 ```
 mkdir build
@@ -154,9 +154,9 @@ VDUSE - на данный момент лучший интерфейс для п
 Чтобы попробовать VDUSE, вам нужно ядро Linux как минимум версии 5.15, собранное с поддержкой
 VDUSE (CONFIG_VDPA=m, CONFIG_VDPA_USER=m, CONFIG_VIRTIO_VDPA=m).

-В ядрах в Debian Linux поддержка пока отключена по умолчанию, так что чтобы попробовать VDUSE
-на Debian, поставьте ядро из Ubuntu [kernel-ppa/mainline](https://kernel.ubuntu.com/~kernel-ppa/mainline/),
-из Proxmox или соберите модули для ядра Debian вручную:
+В ядрах в Debian Linux эти опции включены, только начиная с 6.6, так что установите свежее ядро
+из bookworm-backports, trixie или из более новой версии Debian, если хотите попробовать VDUSE.
+Либо же вы можете самостоятельно собрать модули для установленного ядра:

 ```
 mkdir build
@@ -232,6 +232,7 @@ class EtcdAdapter
    async become_master()
    {
        const state = { ...this.mon.get_mon_state(), id: ''+this.mon.etcd_lease_id };
+        console.log('Waiting to become master');
        // eslint-disable-next-line no-constant-condition
        while (1)
        {
@@ -243,7 +244,6 @@ class EtcdAdapter
            {
                break;
            }
-            console.log('Waiting to become master');
            await new Promise(ok => setTimeout(ok, this.mon.config.etcd_start_timeout));
        }
        console.log('Became master');
@@ -56,6 +56,7 @@ const etcd_tree = {
            osd_out_time: 600, // seconds. min: 0
            placement_levels: { datacenter: 1, rack: 2, host: 3, osd: 4, ... },
            use_old_pg_combinator: false,
+            osd_backfillfull_ratio: 0.99,
            // client and osd
            tcp_header_buffer_size: 65536,
            use_sync_send_recv: false,
@@ -74,6 +74,7 @@ class Mon
        this.state = JSON.parse(JSON.stringify(etcd_tree));
        this.prev_stats = { osd_stats: {}, osd_diff: {} };
        this.recheck_pgs_active = false;
+        this.updating_total_stats = false;
        this.watcher_active = false;
        this.old_pg_config = false;
        this.old_pg_stats_seen = false;
@@ -218,6 +219,10 @@ class Mon
            }
            else if (key != '/stats' && key.substr(0, 13) != '/inode/stats/')
            {
+                if (key == '/config/pgs' && !kv.value)
+                {
+                    this.old_pg_config = false;
+                }
                changed = true;
            }
            if (this.config.verbose)
@@ -422,30 +427,6 @@ class Mon
                    throw new Error('Failed to duplicate old PG config to new PG config');
            }
            this.old_pg_config = true;
-            this.old_pg_config_timer = setInterval(() => this.check_clear_old_config().catch(console.error),
-                this.config.old_pg_config_clear_interval||3600000);
-        }
-    }
-
-    async check_clear_old_config()
-    {
-        if (this.old_pg_config && this.old_pg_stats_seen)
-        {
-            this.old_pg_stats_seen = false;
-            return;
-        }
-        if (this.old_pg_config)
-        {
-            await this.etcd.etcd_call('/kv/txn', { success: [
-                { requestDeleteRange: { key: b64(this.config.etcd_prefix+'/config/pgs') } },
-                { requestDeleteRange: { key: b64(this.config.etcd_prefix+'/pg/stats/'), range_end: b64(this.config.etcd_prefix+'/pg/stats0') } },
-            ] }, this.config.etcd_mon_timeout, this.config.etcd_mon_retries);
-            this.old_pg_config = false;
-        }
-        if (this.old_pg_config_timer)
-        {
-            clearInterval(this.old_pg_config_timer);
-            this.old_pg_config_timer = null;
        }
    }

@@ -587,6 +568,7 @@ class Mon

    async apply_pool_pgs(results, up_osds, osd_tree, tree_hash)
    {
+        const etcd_request = { compare: [], success: [] };
        for (const pool_id in (this.state.pg.config||{}).items||{})
        {
            // We should stop all PGs when deleting a pool or changing its PG count
@@ -599,9 +581,24 @@ class Mon
                    return false;
                }
            }
+            if (!this.state.config.pools[pool_id])
+            {
+                // Delete PG history and stats of the deleted pool
+                etcd_request.success.push({ requestDeleteRange: {
+                    key: b64(this.config.etcd_prefix+'/pg/history/'+pool_id+'/'),
+                    range_end: b64(this.config.etcd_prefix+'/pg/history/'+pool_id+'0'),
+                } });
+                etcd_request.success.push({ requestDeleteRange: {
+                    key: b64(this.config.etcd_prefix+'/pg/stats/'+pool_id+'/'),
+                    range_end: b64(this.config.etcd_prefix+'/pg/stats/'+pool_id+'0'),
+                } });
+                etcd_request.success.push({ requestDeleteRange: {
+                    key: b64(this.config.etcd_prefix+'/pgstats/'+pool_id+'/'),
+                    range_end: b64(this.config.etcd_prefix+'/pgstats/'+pool_id+'0'),
+                } });
+            }
        }
        const new_pg_config = JSON.parse(JSON.stringify(this.state.pg.config));
-        const etcd_request = { compare: [], success: [] };
        for (const pool_id in (new_pg_config||{}).items||{})
        {
            if (!this.state.config.pools[pool_id])
@@ -662,7 +659,19 @@ class Mon
                this.etcd_watch_revision, pool_id, up_osds, osd_tree, real_prev_pgs, pool_res.pgs, pg_history);
        }
        new_pg_config.hash = tree_hash;
-        return await this.save_pg_config(new_pg_config, etcd_request);
+        const { backfillfull_pools, backfillfull_osds } = sum_object_counts(
+            { ...this.state, pg: { ...this.state.pg, config: new_pg_config } }, this.config
+        );
+        if (backfillfull_pools.join(',') != ((this.state.pg.config||{}).backfillfull_pools||[]).join(','))
+        {
+            this.log_backfillfull(backfillfull_osds, backfillfull_pools);
+        }
+        new_pg_config.backfillfull_pools = backfillfull_pools.length ? backfillfull_pools : undefined;
+        if (!await this.save_pg_config(new_pg_config, etcd_request))
+        {
+            return false;
+        }
+        return true;
    }

    async save_pg_config(new_pg_config, etcd_request = { compare: [], success: [] })
@@ -734,7 +743,7 @@ class Mon
    async update_total_stats()
    {
        const txn = [];
-        const { object_counts, object_bytes } = sum_object_counts(this.state, this.config);
+        const { object_counts, object_bytes, backfillfull_pools, backfillfull_osds } = sum_object_counts(this.state, this.config);
        let stats = sum_op_stats(this.state.osd, this.prev_stats);
        let { inode_stats, seen_pools } = sum_inode_stats(this.state, this.prev_stats);
        stats.object_counts = object_counts;
@@ -787,6 +796,27 @@ class Mon
        {
            await this.etcd.etcd_call('/kv/txn', { success: txn }, this.config.etcd_mon_timeout, 0);
        }
+        if (!this.recheck_pgs_active &&
+            backfillfull_pools.join(',') != ((this.state.pg.config||{}).backfillfull_pools||[]).join(','))
+        {
+            this.log_backfillfull(backfillfull_osds, backfillfull_pools);
+            const new_pg_config = { ...this.state.pg.config, backfillfull_pools: backfillfull_pools.length ? backfillfull_pools : undefined };
+            await this.save_pg_config(new_pg_config);
+        }
+    }
+
+    log_backfillfull(osds, pools)
+    {
+        for (const osd in osds)
+        {
+            const bf = osds[osd];
+            console.log('OSD '+osd+' may fill up during rebalance: capacity '+(bf.cap/1024n/1024n)+
+                ' MB, target user data '+(bf.clean/1024n/1024n)+' MB');
+        }
+        console.log(
+            (pools.length ? 'Pool(s) '+pools.join(', ') : 'No pools')+
+            ' are backfillfull now, applying rebalance configuration'
+        );
    }

    schedule_update_stats()
@@ -798,7 +828,21 @@ class Mon
        this.stats_timer = setTimeout(() =>
        {
            this.stats_timer = null;
-            this.update_total_stats().catch(console.error);
+            if (this.updating_total_stats)
+            {
+                this.schedule_update_stats();
+                return;
+            }
+            this.updating_total_stats = true;
+            try
+            {
+                this.update_total_stats().catch(console.error);
+            }
+            catch (e)
+            {
+                console.error(e);
+            }
+            this.updating_total_stats = false;
        }, this.config.mon_stats_timeout);
    }

@@ -1,6 +1,6 @@
 {
  "name": "vitastor-mon",
-  "version": "1.8.0",
+  "version": "1.10.0",
  "description": "Vitastor SDS monitor service",
  "main": "mon-main.js",
  "scripts": {
@@ -109,6 +109,8 @@ function sum_object_counts(state, global_config)
            pgstats[pool_id] = { ...(state.pg.stats[pool_id] || {}), ...(pgstats[pool_id] || {}) };
        }
    }
+    const pool_per_osd = {};
+    const clean_per_osd = {};
    for (const pool_id in pgstats)
    {
        let object_size = 0;
@@ -143,10 +145,45 @@ function sum_object_counts(state, global_config)
                        object_bytes[k] += BigInt(st[k+'_count']) * object_size;
                    }
                }
+                if (st.object_count)
+                {
+                    for (const pg_osd of (((state.pg.config.items||{})[pool_id]||{})[pg_num]||{}).osd_set||[])
+                    {
+                        if (!(pg_osd in clean_per_osd))
+                        {
+                            clean_per_osd[pg_osd] = 0n;
+                        }
+                        clean_per_osd[pg_osd] += BigInt(st.object_count);
+                        pool_per_osd[pg_osd] = pool_per_osd[pg_osd]||{};
+                        pool_per_osd[pg_osd][pool_id] = true;
+                    }
+                }
            }
        }
    }
-    return { object_counts, object_bytes };
+    // If clean_per_osd[osd] is larger than osd capacity then it will fill up during rebalance
+    let backfillfull_pools = {};
+    let backfillfull_osds = {};
+    for (const osd in clean_per_osd)
+    {
+        const st = state.osd.stats[osd];
+        if (!st || !st.size || !st.data_block_size)
+        {
+            continue;
+        }
+        let cap = BigInt(st.size)/BigInt(st.data_block_size);
+        cap = cap * BigInt((global_config.osd_backfillfull_ratio||0.99)*1000000) / 1000000n;
+        if (cap < clean_per_osd[osd])
+        {
+            backfillfull_osds[osd] = { cap: BigInt(st.size), clean: clean_per_osd[osd]*BigInt(st.data_block_size) };
+            for (const pool_id in pool_per_osd[osd])
+            {
+                backfillfull_pools[pool_id] = true;
+            }
+        }
+    }
+    backfillfull_pools = Object.keys(backfillfull_pools).sort();
+    return { object_counts, object_bytes, backfillfull_pools, backfillfull_osds };
 }

 // sum_inode_stats(this.state, this.prev_stats)
@@ -3,7 +3,9 @@
 set -e

 reapply_patch() {
-	if ! patch -f --dry-run -F 0 -R $1 < $2 >/dev/null; then
+	if ! [[ -e $1 ]]; then
+		echo "$1 does not exist, OpenNebula is not installed"
+	elif ! patch -f --dry-run -F 0 -R $1 < $2 >/dev/null; then
 		already_applied=0
 		if ! patch --no-backup-if-mismatch -r - -F 0 -f $1 < $2; then
 			applied_ok=0
@@ -15,8 +17,13 @@ echo "Reapplying Vitastor patches to OpenNebula's oned.conf, vmm_execrc and down
 already_applied=1
 applied_ok=1
 reapply_patch /var/lib/one/remotes/datastore/downloader.sh /var/lib/one/remotes/datastore/vitastor/downloader-vitastor.sh.diff
-reapply_patch /etc/one/oned.conf /var/lib/one/remotes/datastore/vitastor/oned.conf.diff
 reapply_patch /etc/one/vmm_exec/vmm_execrc /var/lib/one/remotes/datastore/vitastor/vmm_execrc.diff
+if [[ -e /etc/one/oned.conf ]]; then
+	if ! /var/lib/one/remotes/datastore/vitastor/patch-oned-conf.py /etc/one/oned.conf; then
+		applied_ok=0
+		already_applied=0
+	fi
+fi
 if [[ "$already_applied" = 1 ]]; then
 	echo "OK: Vitastor OpenNebula patches are already applied"
 elif [[ "$applied_ok" = 1 ]]; then
@@ -54,20 +54,20 @@ index be02d646a8..27f876ec36 100644
 TM_MAD_CONF = [
     NAME = "iscsi_libvirt", LN_TARGET = "NONE", CLONE_TARGET = "SELF", SHARED = "YES",
     DS_MIGRATE = "NO", DRIVER = "raw"
-@@ -1219,9 +1230,16 @@ DS_MAD_CONF = [
+@@ -1263,9 +1274,16 @@ DS_MAD_CONF = [
     NAME = "ceph",
     REQUIRED_ATTRS = "DISK_TYPE,BRIDGE_LIST",
     PERSISTENT_ONLY = "NO",
     MARKETPLACE_ACTIONS = "export"
- ]
- 
+]
+
 +DS_MAD_CONF = [
 +    NAME = "vitastor",
 +    REQUIRED_ATTRS = "DISK_TYPE,BRIDGE_LIST",
 +    PERSISTENT_ONLY = "NO",
 +    MARKETPLACE_ACTIONS = "export"
-+]
-+
+ ]
+ 
 DS_MAD_CONF = [
     NAME = "dev", REQUIRED_ATTRS = "DISK_TYPE", PERSISTENT_ONLY = "YES"
 ]
@@ -0,0 +1,115 @@
+#!/usr/bin/env python3
+# Patch /etc/one/oned.conf for Vitastor support
+# -s = also enable save.vitastor/restore.vitastor overrides
+
+import re
+import os
+import sys
+
+class Fixer:
+    save_restore = 0
+
+    def require_sub_cb(self, m, cb):
+        self.found = 1
+        return cb(m)
+
+    def require_sub(self, regexp, cb, text, error):
+        self.found = 0
+        new_text = re.sub(regexp, lambda m: self.require_sub_cb(m, cb), text)
+        if not self.found and error:
+            self.errors.append(error)
+        return new_text
+
+    def fix(self, oned_conf):
+        self.errors = []
+        self.kvm_found = 0
+        oned_conf = self.require_sub(r'((?:^|\n)[ \t]*VM_MAD\s*=\s*\[)([^\]]+)\]', lambda m: m.group(1)+self.fix_vm_mad(m.group(2))+']', oned_conf, 'VM_MAD not found')
+        if not self.kvm_found:
+            self.errors.append("VM_MAD[NAME=kvm].ARGUMENTS not found")
+        oned_conf = self.require_sub(r'((?:^|\n)[ \t]*TM_MAD\s*=\s*\[)([^\]]+)\]', lambda m: m.group(1)+self.fix_tm_mad(m.group(2))+']', oned_conf, 'TM_MAD not found')
+        oned_conf = self.require_sub(r'((?:^|\n)[ \t]*DATASTORE_MAD\s*=\s*\[)([^\]]+)\]', lambda m: m.group(1)+self.fix_datastore_mad(m.group(2))+']', oned_conf, 'DATASTORE_MAD not found')
+        if oned_conf[-1:] != '\n':
+            oned_conf += '\n'
+        if not re.compile(r'(^|\n)[ \t]*INHERIT_DATASTORE_ATTR\s*=\s*"VITASTOR_CONF"').search(oned_conf):
+            oned_conf += '\nINHERIT_DATASTORE_ATTR="VITASTOR_CONF"\n'
+        if not re.compile(r'(^|\n)[ \t]*INHERIT_DATASTORE_ATTR\s*=\s*"IMAGE_PREFIX"').search(oned_conf):
+            oned_conf += '\nINHERIT_DATASTORE_ATTR="IMAGE_PREFIX"\n'
+        if not re.compile(r'(^|\n)[ \t]*TM_MAD_CONF\s*=\s*\[[^\]]*NAME\s*=\s*"vitastor"').search(oned_conf):
+            oned_conf += ('\nTM_MAD_CONF = [\n'+
+                '    NAME = "vitastor", LN_TARGET = "NONE", CLONE_TARGET = "SELF", SHARED = "YES",\n'+
+                '    DS_MIGRATE = "NO", DRIVER = "raw", ALLOW_ORPHANS="format",\n'+
+                '    TM_MAD_SYSTEM = "ssh,shared", LN_TARGET_SSH = "SYSTEM", CLONE_TARGET_SSH = "SYSTEM",\n'+
+                '    DISK_TYPE_SSH = "FILE", LN_TARGET_SHARED = "NONE",\n'+
+                '    CLONE_TARGET_SHARED = "SELF", DISK_TYPE_SHARED = "FILE"\n'+
+                ']\n')
+        if not re.compile(r'(^|\n)[ \t]*DS_MAD_CONF\s*=\s*\[[^\]]*NAME\s*=\s*"vitastor"').search(oned_conf):
+            oned_conf += ('\nDS_MAD_CONF = [\n'+
+                '    NAME = "vitastor",\n'+
+                '    REQUIRED_ATTRS = "DISK_TYPE,BRIDGE_LIST",\n'+
+                '    PERSISTENT_ONLY = "NO",\n'+
+                '    MARKETPLACE_ACTIONS = "export"\n'+
+                ']\n')
+        return oned_conf
+
+    def fix_vm_mad(self, vm_mad_params):
+        if re.compile(r'\bNAME\s*=\s*"kvm"').search(vm_mad_params):
+            vm_mad_params = re.sub(r'\b(ARGUMENTS\s*=\s*")([^"]+)"', lambda m: m.group(1)+self.fix_vm_mad_args(m.group(2))+'"', vm_mad_params)
+            self.kvm_found = 1
+        return vm_mad_params
+
+    def fix_vm_mad_args(self, args):
+        args = self.fix_vm_mad_override(args, 'deploy')
+        if self.save_restore:
+            args = self.fix_vm_mad_override(args, 'save')
+            args = self.fix_vm_mad_override(args, 'restore')
+        return args
+
+    def fix_vm_mad_override(self, args, override):
+        m = re.compile(r'-l (\S+)').search(args)
+        if m and re.compile(override+'='+override+'.vitastor').search(m.group(1)):
+            return args
+        elif m and re.compile(override+'=').search(m.group(1)):
+            self.errors.append(override+"= is already overridden in -l option in VM_MAD[NAME=kvm].ARGUMENTS")
+            return args
+        elif m:
+            return self.require_sub(r'-l (\S+)', lambda m: '-l '+m.group(1)+','+override+'='+override+'.vitastor', args, '-l option not found in VM_MAD[NAME=kvm].ARGUMENTS')
+        else:
+            return args+' -l '+override+'='+override+'.vitastor'
+
+    def fix_tm_mad(self, params):
+        return self.require_sub(r'\b(ARGUMENTS\s*=\s*")([^"]+)"', lambda m: m.group(1)+self.fix_tm_mad_args('d', m.group(2), "TM_MAD")+'"', params, "TM_MAD.ARGUMENTS not found")
+
+    def fix_tm_mad_args(self, opt, args, v):
+        return self.require_sub('(-'+opt+r') (\S+)', lambda m: self.fix_tm_mad_arg(m), args, "-"+opt+" option not found in "+v+".ARGUMENTS")
+
+    def fix_tm_mad_arg(self, m):
+        a = m.group(2).split(',')
+        if 'vitastor' not in a:
+            a += [ 'vitastor' ]
+        return m.group(1)+' '+(','.join(a))
+
+    def fix_datastore_mad(self, params):
+        params = self.require_sub(r'\b(ARGUMENTS\s*=\s*")([^"]+)"', lambda m: m.group(1)+self.fix_tm_mad_args('d', m.group(2), "DATASTORE_MAD")+'"', params, "DATASTORE_MAD.ARGUMENTS not found")
+        return self.require_sub(r'\b(ARGUMENTS\s*=\s*")([^"]+)"', lambda m: m.group(1)+self.fix_tm_mad_args('s', m.group(2), "DATASTORE_MAD")+'"', params, "")
+
+fixer = Fixer()
+oned_conf_file = ''
+for arg in sys.argv[1:]:
+    if arg == '-s':
+        fixer.save_restore = 1
+    else:
+        oned_conf_file = arg
+        break
+if not oned_conf_file:
+    sys.stderr.write("USAGE: ./patch-oned-conf.py [-s] /etc/one/oned.conf\n-s means also enable save.vitastor/restore.vitastor overrides\n")
+    sys.exit(1)
+with open(oned_conf_file, 'r') as fd:
+    oned_conf = fd.read()
+new_conf = fixer.fix(oned_conf)
+if new_conf != oned_conf:
+    os.rename(oned_conf_file, oned_conf_file+'.bak')
+    with open(oned_conf_file, 'w') as fd:
+        fd.write(new_conf)
+if len(fixer.errors) > 0:
+    sys.stderr.write("ERROR: Failed to patch "+oned_conf_file+", patch it manually. Errors:\n- "+('\n- '.join(fixer.errors))+'\n')
+    sys.exit(1)
@@ -4,12 +4,15 @@
 # Copyright (c) Vitaliy Filippov, 2024+
 # License: Apache-2.0 http://www.apache.org/licenses/LICENSE-2.0

+set -e
+
 DRIVER_PATH=$(dirname $0)
 DEP_FILE=$1
 DEP_FILE_LOCATION=$(dirname $DEP_FILE)
+HOST=$2

 cat > $DEP_FILE

 python3 $DRIVER_PATH/deploy_vitastor.py $DEP_FILE $DEP_FILE_LOCATION/vm.xml

-cat $DEP_FILE | $DRIVER_PATH/deploy $@
+cat "$DEP_FILE" | ssh "$HOST" "'${SCRIPTS_REMOTE_DIR:-/var/tmp/one}/vmm/kvm/deploy' '$DEP_FILE'"
@@ -54,5 +54,4 @@ for disk in dep.findall('./devices/disk[@type="file"]'):
    except Exception as e:
        print("Error: {}".format(e), file=stderr)

-if changed:
-    ET.ElementTree(dep).write(dep_file)
+ET.ElementTree(dep).write(dep_file)
@@ -10,7 +10,7 @@ source $DRIVER_PATH/../../etc/vmm/kvm/kvmrc
 source $DRIVER_PATH/../../scripts_common.sh

 FILE=$1
-HOST=$2
+DST_HOST=$2
 DEPLOY_ID=$3
 VM_ID=$4
 DS_ID=$5
@@ -33,7 +33,8 @@ fi

 SRC_IMAGE="${IMAGE_PREFIX}-sys-${VM_ID}-checkpoint"

-exec_and_log "$CLI dd iimg=$SRC_IMAGE of=$FILE" "Error exporting checkpoint into from $SRC_IMAGE to $FILE"
-exec_and_log "$CLI rm $SRC_IMAGE" "Error removing checkpoint $SRC_IMAGE"
+ssh_exec_and_log "$DST_HOST" "$CLI dd iimg=$SRC_IMAGE of=$FILE" "Error exporting checkpoint into from $SRC_IMAGE to $FILE"
+ssh_exec_and_log "$DST_HOST" "$CLI rm $SRC_IMAGE" "Error removing checkpoint $SRC_IMAGE"

-"$DRIVER_PATH"/restore $@
+set -e
+ssh "$DST_HOST" "'${SCRIPTS_REMOTE_DIR:-/var/tmp/one}/vmm/kvm/restore' $@"
@@ -11,11 +11,14 @@ source $DRIVER_PATH/../../scripts_common.sh

 DEPLOY_ID=$1
 FILE=$2
+SRC_HOST=$3
 VM_ID=$4
 DS_ID=$5

-rm -f "$FILE"
-"$DRIVER_PATH"/save $@
+set -e
+ssh "$SRC_HOST" 'rm -f "$FILE"'
+ssh "$SRC_HOST" "'${SCRIPTS_REMOTE_DIR:-/var/tmp/one}/vmm/kvm/save' $@"
+set +e

 XPATH="${DRIVER_PATH}/../../datastore/xpath.rb --stdin"

@@ -42,7 +45,7 @@ fi

 DST_IMAGE="${IMAGE_PREFIX}-sys-${VM_ID}-checkpoint"

-exec_and_log "$CLI dd if=$FILE oimg=$DST_IMAGE conv=trunc" "Error importing checkpoint into $DST_IMAGE"
-exec_and_log "$RM -f $FILE" "Error removing checkpoint ($FILE)"
+ssh_exec_and_log "$SRC_HOST" "$CLI dd if=$FILE oimg=$DST_IMAGE conv=trunc" "Error importing checkpoint into $DST_IMAGE"
+ssh_exec_and_log "$SRC_HOST" "$RM -f $FILE" "Error removing checkpoint ($FILE)"

 exit 0
@@ -50,7 +50,7 @@ from cinder.volume import configuration
 from cinder.volume import driver
 from cinder.volume import volume_utils

-VITASTOR_VERSION = '1.8.0'
+VITASTOR_VERSION = '1.10.0'

 LOG = logging.getLogger(__name__)

@@ -306,12 +306,12 @@ index e5ff653a60..884ecc79ea 100644
 +        etcd = virBufferContentAndReset(&buf);
 +    }
 +
-+    if (virJSONValueObjectCreate(&ret,
-+                                 "S:etcd-host", etcd,
-+                                 "S:etcd-prefix", src->query,
-+                                 "S:config-path", src->configFile,
-+                                 "s:image", src->path,
-+                                 NULL) < 0)
+    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;
@@ -0,0 +1,193 @@
+Index: pve-qemu-kvm-9.0.0/block/meson.build
+===================================================================
+--- pve-qemu-kvm-9.0.0.orig/block/meson.build
+++ pve-qemu-kvm-9.0.0/block/meson.build
+@@ -126,6 +126,7 @@ 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()
+Index: pve-qemu-kvm-9.0.0/meson.build
+===================================================================
+--- pve-qemu-kvm-9.0.0.orig/meson.build
+++ pve-qemu-kvm-9.0.0/meson.build
+@@ -1452,6 +1452,26 @@ if not get_option('rbd').auto() or have_
+   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
+@@ -2254,6 +2274,7 @@ 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'))
+@@ -4454,6 +4475,7 @@ summary_info += {'fdt support':       fd
+ 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}
+Index: pve-qemu-kvm-9.0.0/meson_options.txt
+===================================================================
+--- pve-qemu-kvm-9.0.0.orig/meson_options.txt
+++ pve-qemu-kvm-9.0.0/meson_options.txt
+@@ -194,6 +194,8 @@ option('lzo', type : 'feature', value :
+        description: 'lzo compression 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',
+Index: pve-qemu-kvm-9.0.0/qapi/block-core.json
+===================================================================
+--- pve-qemu-kvm-9.0.0.orig/qapi/block-core.json
+++ pve-qemu-kvm-9.0.0/qapi/block-core.json
+@@ -3481,7 +3481,7 @@
+             'raw', 'rbd',
+             { 'name': 'replication', 'if': 'CONFIG_REPLICATION' },
+             'pbs',
+-            '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' },
+@@ -4591,6 +4591,28 @@
+             '*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:
+ #
+ # An enumeration of replication modes.
+@@ -5053,6 +5075,7 @@
+       'throttle':   'BlockdevOptionsThrottle',
+       'vdi':        'BlockdevOptionsGenericFormat',
+       'vhdx':       'BlockdevOptionsGenericFormat',
+      'vitastor':   'BlockdevOptionsVitastor',
+       'virtio-blk-vfio-pci':
+                     { 'type': 'BlockdevOptionsVirtioBlkVfioPci',
+                       'if': 'CONFIG_BLKIO' },
+@@ -5498,6 +5521,20 @@
+             '*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:
+ #
+ # Subformat options for VMDK images
+@@ -5719,6 +5753,7 @@
+       'ssh':            'BlockdevCreateOptionsSsh',
+       'vdi':            'BlockdevCreateOptionsVdi',
+       'vhdx':           'BlockdevCreateOptionsVhdx',
+      'vitastor':       'BlockdevCreateOptionsVitastor',
+       'vmdk':           'BlockdevCreateOptionsVmdk',
+       'vpc':            'BlockdevCreateOptionsVpc'
+   } }
+Index: pve-qemu-kvm-9.0.0/scripts/ci/org.centos/stream/8/x86_64/configure
+===================================================================
+--- pve-qemu-kvm-9.0.0.orig/scripts/ci/org.centos/stream/8/x86_64/configure
+++ pve-qemu-kvm-9.0.0/scripts/ci/org.centos/stream/8/x86_64/configure
+@@ -30,7 +30,7 @@
+ --with-suffix="qemu-kvm" \
+ --firmwarepath=/usr/share/qemu-firmware \
+ --target-list="x86_64-softmmu" \
+---block-drv-rw-whitelist="qcow2,raw,file,host_device,nbd,iscsi,rbd,blkdebug,luks,null-co,nvme,copy-on-read,throttle,gluster" \
+--block-drv-rw-whitelist="qcow2,raw,file,host_device,nbd,iscsi,rbd,vitastor,blkdebug,luks,null-co,nvme,copy-on-read,throttle,gluster" \
+ --audio-drv-list="" \
+ --block-drv-ro-whitelist="vmdk,vhdx,vpc,https,ssh" \
+ --with-coroutine=ucontext \
+@@ -176,6 +176,7 @@
+ --enable-opengl \
+ --enable-pie \
+ --enable-rbd \
+--enable-vitastor \
+ --enable-rdma \
+ --enable-seccomp \
+ --enable-snappy \
+Index: pve-qemu-kvm-9.0.0/scripts/meson-buildoptions.sh
+===================================================================
+--- pve-qemu-kvm-9.0.0.orig/scripts/meson-buildoptions.sh
+++ pve-qemu-kvm-9.0.0/scripts/meson-buildoptions.sh
+@@ -168,6 +168,7 @@ meson_options_help() {
+   printf "%s\n" '  qed             qed image format support'
+   printf "%s\n" '  qga-vss         build QGA VSS support (broken with MinGW)'
+   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" '  rutabaga-gfx    rutabaga_gfx support'
+@@ -445,6 +446,8 @@ _meson_option_parse() {
+     --disable-qom-cast-debug) printf "%s" -Dqom_cast_debug=false ;;
+     --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 ;;
@@ -0,0 +1,172 @@
+Index: pve-qemu-kvm-9.1.2/block/meson.build
+===================================================================
+--- pve-qemu-kvm-9.1.2.orig/block/meson.build
+++ pve-qemu-kvm-9.1.2/block/meson.build
+@@ -126,6 +126,7 @@ 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()
+Index: pve-qemu-kvm-9.1.2/meson.build
+===================================================================
+--- pve-qemu-kvm-9.1.2.orig/meson.build
+++ pve-qemu-kvm-9.1.2/meson.build
+@@ -1516,6 +1516,26 @@ if not get_option('rbd').auto() or have_
+   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
+@@ -2367,6 +2387,7 @@ 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'))
+@@ -4534,6 +4555,7 @@ summary_info += {'fdt support':       fd
+ 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}
+Index: pve-qemu-kvm-9.1.2/meson_options.txt
+===================================================================
+--- pve-qemu-kvm-9.1.2.orig/meson_options.txt
+++ pve-qemu-kvm-9.1.2/meson_options.txt
+@@ -194,6 +194,8 @@ option('lzo', type : 'feature', value :
+        description: 'lzo compression 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',
+Index: pve-qemu-kvm-9.1.2/qapi/block-core.json
+===================================================================
+--- pve-qemu-kvm-9.1.2.orig/qapi/block-core.json
+++ pve-qemu-kvm-9.1.2/qapi/block-core.json
+@@ -3477,7 +3477,7 @@
+             'raw', 'rbd',
+             { 'name': 'replication', 'if': 'CONFIG_REPLICATION' },
+             'pbs',
+-            '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' },
+@@ -4588,6 +4588,28 @@
+             '*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:
+ #
+ # An enumeration of replication modes.
+@@ -5050,6 +5072,7 @@
+       'throttle':   'BlockdevOptionsThrottle',
+       'vdi':        'BlockdevOptionsGenericFormat',
+       'vhdx':       'BlockdevOptionsGenericFormat',
+      'vitastor':   'BlockdevOptionsVitastor',
+       'virtio-blk-vfio-pci':
+                     { 'type': 'BlockdevOptionsVirtioBlkVfioPci',
+                       'if': 'CONFIG_BLKIO' },
+@@ -5497,6 +5520,20 @@
+             '*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:
+ #
+ # Subformat options for VMDK images
+@@ -5718,6 +5755,7 @@
+       'ssh':            'BlockdevCreateOptionsSsh',
+       'vdi':            'BlockdevCreateOptionsVdi',
+       'vhdx':           'BlockdevCreateOptionsVhdx',
+      'vitastor':       'BlockdevCreateOptionsVitastor',
+       'vmdk':           'BlockdevCreateOptionsVmdk',
+       'vpc':            'BlockdevCreateOptionsVpc'
+   } }
+Index: pve-qemu-kvm-9.1.2/scripts/meson-buildoptions.sh
+===================================================================
+--- pve-qemu-kvm-9.1.2.orig/scripts/meson-buildoptions.sh
+++ pve-qemu-kvm-9.1.2/scripts/meson-buildoptions.sh
+@@ -168,6 +168,7 @@ 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" '  rutabaga-gfx    rutabaga_gfx support'
+@@ -444,6 +445,8 @@ _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 ;;
@@ -0,0 +1,172 @@
+diff --git a/block/meson.build b/block/meson.build
+index f1262ec2ba..3cf3e23f16 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 fbda17c987..3edac22aff 100644
+--- a/meson.build
+++ b/meson.build
+@@ -1510,6 +1510,26 @@ 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
+@@ -2351,6 +2371,7 @@ 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'))
+@@ -4510,6 +4531,7 @@ 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 0269fa0f16..4740ffdc27 100644
+--- a/meson_options.txt
+++ b/meson_options.txt
+@@ -194,6 +194,8 @@ option('lzo', type : 'feature', value : 'auto',
+        description: 'lzo compression 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 aa40d44f1d..bbee6a0e9c 100644
+--- a/qapi/block-core.json
+++ b/qapi/block-core.json
+@@ -3203,7 +3203,7 @@
+             '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' },
+@@ -4286,6 +4286,28 @@
+             '*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:
+ #
+@@ -4742,6 +4764,7 @@
+       'throttle':   'BlockdevOptionsThrottle',
+       'vdi':        'BlockdevOptionsGenericFormat',
+       'vhdx':       'BlockdevOptionsGenericFormat',
+      'vitastor':   'BlockdevOptionsVitastor',
+       'virtio-blk-vfio-pci':
+                     { 'type': 'BlockdevOptionsVirtioBlkVfioPci',
+                       'if': 'CONFIG_BLKIO' },
+@@ -5183,6 +5206,20 @@
+             '*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:
+ #
+@@ -5405,6 +5442,7 @@
+       '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 c97079a38c..4623f552ec 100644
+--- a/scripts/meson-buildoptions.sh
+++ b/scripts/meson-buildoptions.sh
+@@ -168,6 +168,7 @@ 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" '  rutabaga-gfx    rutabaga_gfx support'
+@@ -444,6 +445,8 @@ _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 ;;
@@ -1,11 +1,11 @@
 Name:           vitastor
-Version:        1.8.0
+Version:        1.10.0
 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-1.8.0.el7.tar.gz
+Source0:        vitastor-1.10.0.el7.tar.gz

 BuildRequires:  liburing-devel >= 0.6
 BuildRequires:  gperftools-devel
@@ -15,7 +15,7 @@ BuildRequires:  rh-nodejs12-npm
 BuildRequires:  jerasure-devel
 BuildRequires:  libisa-l-devel
 BuildRequires:  gf-complete-devel
-BuildRequires:  libibverbs-devel
+BuildRequires:  rdma-core-devel
 BuildRequires:  cmake3
 BuildRequires:  libnl3-devel
 Requires:       vitastor-osd = %{version}-%{release}
@@ -202,4 +202,8 @@ chown vitastor:vitastor /var/lib/vitastor
 /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
@@ -1,11 +1,11 @@
 Name:           vitastor
-Version:        1.8.0
+Version:        1.10.0
 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-1.8.0.el8.tar.gz
+Source0:        vitastor-1.10.0.el8.tar.gz

 BuildRequires:  liburing-devel >= 0.6
 BuildRequires:  gperftools-devel
@@ -14,7 +14,7 @@ BuildRequires:  nodejs >= 10
 BuildRequires:  jerasure-devel
 BuildRequires:  libisa-l-devel
 BuildRequires:  gf-complete-devel
-BuildRequires:  libibverbs-devel
+BuildRequires:  rdma-core-devel
 BuildRequires:  cmake
 BuildRequires:  libnl3-devel
 Requires:       vitastor-osd = %{version}-%{release}
@@ -199,4 +199,8 @@ chown vitastor:vitastor /var/lib/vitastor
 /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
@@ -1,11 +1,11 @@
 Name:           vitastor
-Version:        1.8.0
+Version:        1.10.0
 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-1.8.0.el9.tar.gz
+Source0:        vitastor-1.10.0.el9.tar.gz

 BuildRequires:  liburing-devel >= 0.6
 BuildRequires:  gperftools-devel
@@ -74,7 +74,7 @@ Vitastor library headers for development.
 Summary:        Vitastor - fio drivers
 Group:          Development/Libraries
 Requires:       vitastor-client = %{version}-%{release}
-Requires:       fio = 3.27-8.el9
+Requires:       fio = 3.35-1.el9


 %description -n vitastor-fio
@@ -192,4 +192,8 @@ chown vitastor:vitastor /var/lib/vitastor
 /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:]].*$!!g')
+
+
 %changelog
@@ -19,7 +19,7 @@ if("${CMAKE_INSTALL_PREFIX}" MATCHES "^/usr/local/?$")
 	set(CMAKE_INSTALL_RPATH "${CMAKE_INSTALL_PREFIX}/${CMAKE_INSTALL_LIBDIR}")
 endif()

-add_definitions(-DVITASTOR_VERSION="1.8.0")
+add_definitions(-DVITASTOR_VERSION="1.10.0")
 add_definitions(-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 -I ${CMAKE_SOURCE_DIR}/src)
 add_link_options(-fno-omit-frame-pointer)
 if (${WITH_ASAN})
@@ -61,6 +61,10 @@ pkg_check_modules(ISAL libisal)
 if (ISAL_LIBRARIES)
 	add_definitions(-DWITH_ISAL)
 endif (ISAL_LIBRARIES)
+pkg_check_modules(RDMACM librdmacm)
+if (RDMACM_LIBRARIES)
+	add_definitions(-DWITH_RDMACM)
+endif (RDMACM_LIBRARIES)

 add_custom_target(build_tests)
 add_custom_target(test
@@ -993,7 +993,8 @@ int blockstore_impl_t::read_bitmap(object_id oid, uint64_t target_version, void
    {
        while (dirty_it->first.oid == oid)
        {
-            if (target_version >= dirty_it->first.version)
+            // Condition has to be the same as in dequeue_read()
+            if (!IS_IN_FLIGHT(dirty_it->second.state) && target_version >= dirty_it->first.version)
            {
                if (result_version)
                    *result_version = dirty_it->first.version;
@@ -10,7 +10,7 @@ endif (IBVERBS_LIBRARIES)
 add_library(vitastor_common STATIC
 	../util/epoll_manager.cpp etcd_state_client.cpp messenger.cpp ../util/addr_util.cpp
 	msgr_stop.cpp msgr_op.cpp msgr_send.cpp msgr_receive.cpp ../util/ringloop.cpp ../../json11/json11.cpp
-	http_client.cpp osd_ops.cpp pg_states.cpp ../util/timerfd_manager.cpp ../util/str_util.cpp ${MSGR_RDMA}
+	http_client.cpp osd_ops.cpp pg_states.cpp ../util/timerfd_manager.cpp ../util/str_util.cpp ../util/json_util.cpp ${MSGR_RDMA}
 )
 target_link_libraries(vitastor_common pthread)
 target_compile_options(vitastor_common PUBLIC -fPIC)
@@ -88,7 +88,7 @@ add_executable(test_cluster_client
 	EXCLUDE_FROM_ALL
 	../test/test_cluster_client.cpp
 	pg_states.cpp osd_ops.cpp cluster_client.cpp cluster_client_list.cpp cluster_client_wb.cpp msgr_op.cpp ../test/mock/messenger.cpp msgr_stop.cpp
-	etcd_state_client.cpp ../util/timerfd_manager.cpp ../util/str_util.cpp ../../json11/json11.cpp
+	etcd_state_client.cpp ../util/timerfd_manager.cpp ../util/str_util.cpp ../util/json_util.cpp ../../json11/json11.cpp
 )
 target_compile_definitions(test_cluster_client PUBLIC -D__MOCK__)
 target_include_directories(test_cluster_client BEFORE PUBLIC ${CMAKE_SOURCE_DIR}/src/test/mock)
@@ -4,7 +4,7 @@
 #include <stdexcept>
 #include <assert.h>
 #include "cluster_client_impl.h"
-#include "http_client.h" // json_is_true
+#include "json_util.h"

 cluster_client_t::cluster_client_t(ring_loop_t *ringloop, timerfd_manager_t *tfd, json11::Json config)
 {
@@ -955,7 +955,7 @@ void cluster_client_t::slice_rw(cluster_op_t *op)
            ? (stripe + pg_block_size) : (op->offset + op->len);
        op->parts[i].iov.reset();
        op->parts[i].flags = 0;
-        if (op->cur_inode != op->inode || op->opcode == OSD_OP_READ && dirty_copied)
+        if (op->opcode != OSD_OP_READ_CHAIN_BITMAP && op->cur_inode != op->inode || op->opcode == OSD_OP_READ && dirty_copied)
        {
            // Read remaining parts from upper layers
            uint64_t prev = begin, cur = begin;
@@ -176,7 +176,7 @@ void etcd_state_client_t::add_etcd_url(std::string addr)
            exit(1);
        }
        if (!local_ips.size())
-            local_ips = getifaddr_list();
+            local_ips = getifaddr_list(std::vector<std::string>(), true);
        std::string check_addr;
        int pos = addr.find('/');
        int pos2 = addr.find(':');
@@ -785,7 +785,7 @@ void etcd_state_client_t::parse_state(const etcd_kv_t & kv)
        }
        for (auto & pool_item: value.object_items())
        {
-            pool_config_t pc;
+            pool_config_t pc = {};
            // ID
            pool_id_t pool_id;
            char null_byte = 0;
@@ -931,12 +931,28 @@ void etcd_state_client_t::parse_state(const etcd_kv_t & kv)
            // Ignore old key if the new one is present
            return;
        }
+        for (auto & pool_id_json: value["backfillfull_pools"].array_items())
+        {
+            auto pool_id = pool_id_json.uint64_value();
+            auto pool_it = this->pool_config.find(pool_id);
+            if (pool_it != this->pool_config.end())
+            {
+                pool_it->second.backfillfull |= 2;
+            }
+        }
        for (auto & pool_item: this->pool_config)
        {
            for (auto & pg_item: pool_item.second.pg_config)
            {
                pg_item.second.config_exists = false;
            }
+            // 3 = was 1 and became 1, 0 = was 0 and became 0
+            if (pool_item.second.backfillfull == 2 || pool_item.second.backfillfull == 1)
+            {
+                if (on_change_backfillfull_hook)
+                    on_change_backfillfull_hook(pool_item.first);
+            }
+            pool_item.second.backfillfull = pool_item.second.backfillfull >> 1;
        }
        for (auto & pool_item: value["items"].object_items())
        {
@@ -62,6 +62,7 @@ struct pool_config_t
    std::map<pg_num_t, pg_config_t> pg_config;
    uint64_t scrub_interval;
    std::string used_for_fs;
+    int backfillfull;
 };

 struct inode_config_t
@@ -131,6 +132,7 @@ public:
    std::function<json11::Json()> load_pgs_checks_hook;
    std::function<void(bool)> on_load_pgs_hook;
    std::function<void()> on_change_pool_config_hook;
+    std::function<void(pool_id_t)> on_change_backfillfull_hook;
    std::function<void(pool_id_t, pg_num_t, osd_num_t)> on_change_pg_state_hook;
    std::function<void(pool_id_t, pg_num_t)> on_change_pg_history_hook;
    std::function<void(osd_num_t)> on_change_osd_state_hook;
@@ -15,6 +15,7 @@

 #include "addr_util.h"
 #include "str_util.h"
+#include "json_util.h"
 #include "json11/json11.hpp"
 #include "http_client.h"
 #include "timerfd_manager.h"
@@ -61,6 +62,7 @@ struct http_co_t
    inline void end() { ended = true; if (!onstack) { delete this; } }
    void run_cb_and_clear();
    void start_connection();
+    void start_ws_connection();
    void close_connection();
    void next_request();
    void handle_events();
@@ -111,7 +113,7 @@ http_co_t* open_websocket(timerfd_manager_t *tfd, const std::string & host, cons
    handler->keepalive = false;
    handler->request = request;
    handler->response_callback = response_callback;
-    handler->start_connection();
+    handler->start_ws_connection();
    return handler;
 }

@@ -281,6 +283,27 @@ void http_co_t::close_connection()
    epoll_events = 0;
 }

+void http_co_t::start_ws_connection()
+{
+    stackin();
+    start_connection();
+    if (request_timeout > 0)
+    {
+        timeout_id = tfd->set_timer(request_timeout, false, [this](int timer_id)
+        {
+            stackin();
+            if (state != HTTP_CO_WEBSOCKET)
+            {
+                close_connection();
+                parsed = { .error = "Websocket connection timed out" };
+                run_cb_and_clear();
+            }
+            stackout();
+        });
+    }
+    stackout();
+}
+
 void http_co_t::start_connection()
 {
    stackin();
@@ -724,22 +747,3 @@ static bool ws_parse_frame(std::string & buf, int & type, std::string & res)
    buf = buf.substr(hdr+len);
    return true;
 }
-
-// FIXME: move to utils
-bool json_is_true(const json11::Json & val)
-{
-    if (val.is_string())
-        return val == "true" || val == "yes" || val == "1";
-    return val.bool_value();
-}
-
-bool json_is_false(const json11::Json & val)
-{
-    if (val.is_string())
-        return val.string_value() == "false" || val.string_value() == "no" || val.string_value() == "0";
-    if (val.is_number())
-        return val.number_value() == 0;
-    if (val.is_bool())
-        return !val.bool_value();
-    return false;
-}
@@ -48,9 +48,3 @@ void http_request(http_co_t *handler, const std::string & host, const std::strin
    const http_options_t & options, std::function<void(const http_response_t *response)> response_callback);
 void http_post_message(http_co_t *handler, int type, const std::string & msg);
 void http_close(http_co_t *co);
-
-// Utils
-std::string strtolower(const std::string & in);
-// FIXME: move to json11
-bool json_is_true(const json11::Json & val);
-bool json_is_false(const json11::Json & val);
@@ -121,7 +121,7 @@ void osd_messenger_t::init()
    if (use_rdma)
    {
        rdma_context = msgr_rdma_context_t::create(
-            rdma_device != "" ? rdma_device.c_str() : NULL,
+            osd_networks, rdma_device != "" ? rdma_device.c_str() : NULL,
            rdma_port_num, rdma_gid_index, rdma_mtu, rdma_odp, log_level
        );
        if (!rdma_context)
@@ -266,7 +266,8 @@ void osd_messenger_t::parse_config(const json11::Json & config)
    this->rdma_port_num = (uint8_t)config["rdma_port_num"].uint64_value();
    if (!this->rdma_port_num)
        this->rdma_port_num = 1;
-    this->rdma_gid_index = (uint8_t)config["rdma_gid_index"].uint64_value();
+    if (!config["rdma_gid_index"].is_null())
+        this->rdma_gid_index = (uint8_t)config["rdma_gid_index"].uint64_value();
    this->rdma_mtu = (uint32_t)config["rdma_mtu"].uint64_value();
    this->rdma_max_sge = config["rdma_max_sge"].uint64_value();
    if (!this->rdma_max_sge)
@@ -281,6 +282,15 @@ void osd_messenger_t::parse_config(const json11::Json & config)
    if (!this->rdma_max_msg || this->rdma_max_msg > 128*1024*1024)
        this->rdma_max_msg = 129*1024;
    this->rdma_odp = config["rdma_odp"].bool_value();
+    std::vector<std::string> mask;
+    if (config["bind_address"].is_string())
+        mask.push_back(config["bind_address"].string_value());
+    else if (config["osd_network"].is_string())
+        mask.push_back(config["osd_network"].string_value());
+    else
+        for (auto v: config["osd_network"].array_items())
+            mask.push_back(v.string_value());
+    this->osd_networks = mask;
 #endif
    if (!osd_num)
        this->iothread_count = (uint32_t)config["client_iothread_count"].uint64_value();
@@ -165,8 +165,10 @@ protected:

 #ifdef WITH_RDMA
    bool use_rdma = true;
+    std::vector<std::string> osd_networks;
    std::string rdma_device;
-    uint64_t rdma_port_num = 1, rdma_gid_index = 0, rdma_mtu = 0;
+    uint64_t rdma_port_num = 1, rdma_mtu = 0;
+    int rdma_gid_index = -1;
    msgr_rdma_context_t *rdma_context = NULL;
    uint64_t rdma_max_sge = 0, rdma_max_send = 0, rdma_max_recv = 0;
    uint64_t rdma_max_msg = 0;
@@ -177,7 +179,7 @@ protected:
    std::vector<int> read_ready_clients;
    std::vector<int> write_ready_clients;
    // We don't use ringloop->set_immediate here because we may have no ringloop in client :)
-    std::vector<std::function<void()>> set_immediate;
+    std::vector<osd_op_t*> set_immediate_ops;

 public:
    timerfd_manager_t *tfd;
@@ -237,6 +239,8 @@ protected:
    void handle_op_hdr(osd_client_t *cl);
    bool handle_reply_hdr(osd_client_t *cl);
    void handle_reply_ready(osd_op_t *op);
+    void handle_immediate_ops();
+    void clear_immediate_ops(int peer_fd);

 #ifdef WITH_RDMA
    void try_send_rdma(osd_client_t *cl);
@@ -3,6 +3,7 @@

 #include <stdio.h>
 #include <stdlib.h>
+#include "addr_util.h"
 #include "msgr_rdma.h"
 #include "messenger.h"

@@ -69,7 +70,138 @@ msgr_rdma_connection_t::~msgr_rdma_connection_t()
    send_out_size = 0;
 }

-msgr_rdma_context_t *msgr_rdma_context_t::create(const char *ib_devname, uint8_t ib_port, uint8_t gid_index, uint32_t mtu, bool odp, int log_level)
+#ifdef IBV_ADVISE_MR_ADVICE_PREFETCH_NO_FAULT
+static bool is_ipv4_gid(ibv_gid_entry *gidx)
+{
+    return (((uint64_t*)gidx->gid.raw)[0] == 0 &&
+        ((uint32_t*)gidx->gid.raw)[2] == 0xffff0000);
+}
+
+static bool match_gid(ibv_gid_entry *gidx, addr_mask_t *networks, int nnet)
+{
+    if (gidx->gid_type != IBV_GID_TYPE_ROCE_V1 &&
+        gidx->gid_type != IBV_GID_TYPE_ROCE_V2 ||
+        ((uint64_t*)gidx->gid.raw)[0] == 0 &&
+        ((uint64_t*)gidx->gid.raw)[1] == 0)
+    {
+        return false;
+    }
+    if (is_ipv4_gid(gidx))
+    {
+        for (int i = 0; i < nnet; i++)
+        {
+            if (networks[i].family == AF_INET && cidr_match(*(in_addr*)(gidx->gid.raw+12), networks[i].ipv4, networks[i].bits))
+                return true;
+        }
+    }
+    else
+    {
+        for (int i = 0; i < nnet; i++)
+        {
+            if (networks[i].family == AF_INET6 && cidr6_match(*(in6_addr*)gidx->gid.raw, networks[i].ipv6, networks[i].bits))
+                return true;
+        }
+    }
+    return false;
+}
+
+struct matched_dev
+{
+    int dev = -1;
+    int port = -1;
+    int gid = -1;
+    bool rocev2 = false;
+};
+
+static void log_rdma_dev_port_gid(ibv_device *dev, int ib_port, int gid_index, ibv_gid_entry & gidx)
+{
+    bool is4 = ((uint64_t*)gidx.gid.raw)[0] == 0 && ((uint32_t*)gidx.gid.raw)[2] == 0xffff0000;
+    char buf[256];
+    inet_ntop(is4 ? AF_INET : AF_INET6, is4 ? gidx.gid.raw+12 : gidx.gid.raw, buf, sizeof(buf));
+    fprintf(
+        stderr, "Auto-selected RDMA device %s port %d GID %d - ROCEv%d IPv%d %s\n",
+        ibv_get_device_name(dev), ib_port, gid_index,
+        gidx.gid_type == IBV_GID_TYPE_ROCE_V2 ? 2 : 1, is4 ? 4 : 6, buf
+    );
+}
+
+static matched_dev match_device(ibv_device **dev_list, addr_mask_t *networks, int nnet, int log_level)
+{
+    matched_dev best;
+    ibv_device_attr attr;
+    ibv_port_attr portinfo;
+    ibv_gid_entry best_gidx;
+    int res;
+    bool have_non_roce = false, have_roce = false;
+    for (int i = 0; dev_list[i]; ++i)
+    {
+        auto dev = dev_list[i];
+        ibv_context *context = ibv_open_device(dev_list[i]);
+        if ((res = ibv_query_device(context, &attr)) != 0)
+        {
+            fprintf(stderr, "Couldn't query RDMA device %s for its features: %s\n", ibv_get_device_name(dev_list[i]), strerror(res));
+            goto cleanup;
+        }
+        for (int j = 1; j <= attr.phys_port_cnt; j++)
+        {
+            // Try to find a port with matching address
+            if ((res = ibv_query_port(context, j, &portinfo)) != 0)
+            {
+                fprintf(stderr, "Couldn't get RDMA device %s port %d info: %s\n", ibv_get_device_name(dev), j, strerror(res));
+                goto cleanup;
+            }
+            for (int k = 0; k < portinfo.gid_tbl_len; k++)
+            {
+                ibv_gid_entry gidx;
+                if ((res = ibv_query_gid_ex(context, j, k, &gidx, 0)) != 0)
+                {
+                    if (res != ENODATA)
+                    {
+                        fprintf(stderr, "Couldn't read RDMA device %s GID index %d: %s\n", ibv_get_device_name(dev), k, strerror(res));
+                        goto cleanup;
+                    }
+                    else
+                        break;
+                }
+                if (gidx.gid_type != IBV_GID_TYPE_ROCE_V1 &&
+                    gidx.gid_type != IBV_GID_TYPE_ROCE_V2)
+                    have_non_roce = true;
+                else
+                    have_roce = true;
+                if (match_gid(&gidx, networks, nnet))
+                {
+                    // Prefer RoCEv2
+                    if (!best.rocev2)
+                    {
+                        best.dev = i;
+                        best.port = j;
+                        best.gid = k;
+                        best.rocev2 = (gidx.gid_type == IBV_GID_TYPE_ROCE_V2);
+                        best_gidx = gidx;
+                    }
+                }
+            }
+        }
+cleanup:
+        ibv_close_device(context);
+        if (best.rocev2)
+        {
+            break;
+        }
+    }
+    if (best.dev >= 0 && log_level > 0)
+    {
+        log_rdma_dev_port_gid(dev_list[best.dev], best.port, best.gid, best_gidx);
+    }
+    if (best.dev < 0 && have_non_roce && !have_roce)
+    {
+        best.dev = -2;
+    }
+    return best;
+}
+#endif
+
+msgr_rdma_context_t *msgr_rdma_context_t::create(std::vector<std::string> osd_networks, const char *ib_devname, uint8_t ib_port, int gid_index, uint32_t mtu, bool odp, int log_level)
 {
    int res;
    ibv_device **dev_list = NULL;
@@ -80,28 +212,23 @@ msgr_rdma_context_t *msgr_rdma_context_t::create(const char *ib_devname, uint8_t
    clock_gettime(CLOCK_REALTIME, &tv);
    srand48(tv.tv_sec*1000000000 + tv.tv_nsec);
    dev_list = ibv_get_device_list(NULL);
-    if (!dev_list)
+    if (!dev_list || !*dev_list)
    {
        if (errno == -ENOSYS || errno == ENOSYS)
        {
            if (log_level > 0)
                fprintf(stderr, "No RDMA devices found (RDMA device list returned ENOSYS)\n");
        }
+        else if (!*dev_list)
+        {
+            if (log_level > 0)
+                fprintf(stderr, "No RDMA devices found\n");
+        }
        else
            fprintf(stderr, "Failed to get RDMA device list: %s\n", strerror(errno));
        goto cleanup;
    }
-    if (!ib_devname)
-    {
-        ctx->dev = *dev_list;
-        if (!ctx->dev)
-        {
-            if (log_level > 0)
-                fprintf(stderr, "No RDMA devices found\n");
-            goto cleanup;
-        }
-    }
-    else
+    if (ib_devname)
    {
        int i;
        for (i = 0; dev_list[i]; ++i)
@@ -114,6 +241,39 @@ msgr_rdma_context_t *msgr_rdma_context_t::create(const char *ib_devname, uint8_t
            goto cleanup;
        }
    }
+#ifdef IBV_ADVISE_MR_ADVICE_PREFETCH_NO_FAULT
+    else if (osd_networks.size())
+    {
+        std::vector<addr_mask_t> nets;
+        for (auto & netstr: osd_networks)
+        {
+            nets.push_back(cidr_parse(netstr));
+        }
+        auto best = match_device(dev_list, nets.data(), nets.size(), log_level);
+        if (best.dev == -2)
+        {
+            best.dev = 0;
+            if (log_level > 0)
+                fprintf(stderr, "No RoCE devices found, using first available RDMA device %s\n", ibv_get_device_name(*dev_list));
+        }
+        else if (best.dev < 0)
+        {
+            if (log_level > 0)
+                fprintf(stderr, "RDMA device matching osd_network is not found, disabling RDMA\n");
+            goto cleanup;
+        }
+        else
+        {
+            ib_port = best.port;
+            gid_index = best.gid;
+        }
+        ctx->dev = dev_list[best.dev];
+    }
+#endif
+    else
+    {
+        ctx->dev = *dev_list;
+    }

    ctx->context = ibv_open_device(ctx->dev);
    if (!ctx->context)
@@ -123,7 +283,6 @@ msgr_rdma_context_t *msgr_rdma_context_t::create(const char *ib_devname, uint8_t
    }

    ctx->ib_port = ib_port;
-    ctx->gid_index = gid_index;
    if ((res = ibv_query_port(ctx->context, ib_port, &ctx->portinfo)) != 0)
    {
        fprintf(stderr, "Couldn't get RDMA device %s port %d info: %s\n", ibv_get_device_name(ctx->dev), ib_port, strerror(res));
@@ -135,11 +294,55 @@ msgr_rdma_context_t *msgr_rdma_context_t::create(const char *ib_devname, uint8_t
        fprintf(stderr, "RDMA device %s must have local LID because it's not Ethernet, but LID is zero\n", ibv_get_device_name(ctx->dev));
        goto cleanup;
    }
-    if (ibv_query_gid(ctx->context, ib_port, gid_index, &ctx->my_gid))
+
+#ifdef IBV_ADVISE_MR_ADVICE_PREFETCH_NO_FAULT
+    if (gid_index != -1)
+#endif
    {
-        fprintf(stderr, "Couldn't read RDMA device %s GID index %d\n", ibv_get_device_name(ctx->dev), gid_index);
-        goto cleanup;
+        ctx->gid_index = gid_index < 0 ? 0 : gid_index;
+        if (ibv_query_gid(ctx->context, ib_port, gid_index, &ctx->my_gid))
+        {
+            fprintf(stderr, "Couldn't read RDMA device %s GID index %d\n", ibv_get_device_name(ctx->dev), gid_index);
+            goto cleanup;
+        }
    }
+#ifdef IBV_ADVISE_MR_ADVICE_PREFETCH_NO_FAULT
+    else
+    {
+        // Auto-guess GID
+        ibv_gid_entry best_gidx;
+        for (int k = 0; k < ctx->portinfo.gid_tbl_len; k++)
+        {
+            ibv_gid_entry gidx;
+            if (ibv_query_gid_ex(ctx->context, ib_port, k, &gidx, 0) != 0)
+            {
+                fprintf(stderr, "Couldn't read RDMA device %s GID index %d\n", ibv_get_device_name(ctx->dev), k);
+                goto cleanup;
+            }
+            // Skip empty GID
+            if (((uint64_t*)gidx.gid.raw)[0] == 0 &&
+                ((uint64_t*)gidx.gid.raw)[1] == 0)
+            {
+                continue;
+            }
+            // Prefer IPv4 RoCEv2 -> IPv6 RoCEv2 -> IPv4 RoCEv1 -> IPv6 RoCEv1 -> IB
+            if (gid_index == -1 ||
+                gidx.gid_type == IBV_GID_TYPE_ROCE_V2 && best_gidx.gid_type != IBV_GID_TYPE_ROCE_V2 ||
+                gidx.gid_type == IBV_GID_TYPE_ROCE_V1 && best_gidx.gid_type == IBV_GID_TYPE_IB ||
+                gidx.gid_type == best_gidx.gid_type && is_ipv4_gid(&gidx))
+            {
+                gid_index = k;
+                best_gidx = gidx;
+            }
+        }
+        ctx->gid_index = gid_index = (gid_index == -1 ? 0 : gid_index);
+        if (log_level > 0)
+        {
+            log_rdma_dev_port_gid(ctx->dev, ctx->ib_port, ctx->gid_index, best_gidx);
+        }
+        ctx->my_gid = best_gidx.gid;
+    }
+#endif

    ctx->pd = ibv_alloc_pd(ctx->context);
    if (!ctx->pd)
@@ -398,7 +601,7 @@ static int try_send_rdma_copy(osd_client_t *cl, uint8_t *dst, int dst_len)
        iovec & iov = cl->send_list[rc->send_pos];
        uint32_t len = (uint32_t)(iov.iov_len-rc->send_buf_pos < dst_len
            ? iov.iov_len-rc->send_buf_pos : dst_len);
-        memcpy(dst, iov.iov_base+rc->send_buf_pos, len);
+        memcpy(dst, (uint8_t*)iov.iov_base+rc->send_buf_pos, len);
        dst += len;
        dst_len -= len;
        rc->send_buf_pos += len;
@@ -598,6 +801,7 @@ void osd_messenger_t::handle_rdma_events()
                }
                fprintf(stderr, " with status: %s, stopping client\n", ibv_wc_status_str(wc[i].status));
                stop_client(client_id);
+                clear_immediate_ops(client_id);
                continue;
            }
            if (!is_send)
@@ -606,6 +810,7 @@ void osd_messenger_t::handle_rdma_events()
                if (!handle_read_buffer(cl, rc->recv_buffers[rc->next_recv_buf].buf, wc[i].byte_len))
                {
                    // handle_read_buffer may stop the client
+                    clear_immediate_ops(client_id);
                    continue;
                }
                try_recv_rdma_wr(cl, rc->recv_buffers[rc->next_recv_buf]);
@@ -666,9 +871,5 @@ void osd_messenger_t::handle_rdma_events()
            }
        }
    } while (event_count > 0);
-    for (auto cb: set_immediate)
-    {
-        cb();
-    }
-    set_immediate.clear();
+    handle_immediate_ops();
 }
@@ -36,7 +36,7 @@ struct msgr_rdma_context_t
    int max_cqe = 0;
    int used_max_cqe = 0;

-    static msgr_rdma_context_t *create(const char *ib_devname, uint8_t ib_port, uint8_t gid_index, uint32_t mtu, bool odp, int log_level);
+    static msgr_rdma_context_t *create(std::vector<std::string> osd_networks, const char *ib_devname, uint8_t ib_port, int gid_index, uint32_t mtu, bool odp, int log_level);
    ~msgr_rdma_context_t();
 };

@@ -65,6 +65,7 @@ void osd_messenger_t::read_requests()
 bool osd_messenger_t::handle_read(int result, osd_client_t *cl)
 {
    bool ret = false;
+    int peer_fd = cl->peer_fd;
    cl->read_msg.msg_iovlen = 0;
    cl->refs--;
    if (cl->peer_state == PEER_STOPPED)
@@ -101,7 +102,8 @@ bool osd_messenger_t::handle_read(int result, osd_client_t *cl)
        {
            if (!handle_read_buffer(cl, cl->in_buf, result))
            {
-                goto fin;
+                clear_immediate_ops(peer_fd);
+                return false;
            }
        }
        else
@@ -113,7 +115,8 @@ bool osd_messenger_t::handle_read(int result, osd_client_t *cl)
            {
                if (!handle_finished_read(cl))
                {
-                    goto fin;
+                    clear_immediate_ops(peer_fd);
+                    return false;
                }
            }
        }
@@ -122,15 +125,47 @@ bool osd_messenger_t::handle_read(int result, osd_client_t *cl)
            ret = true;
        }
    }
-fin:
-    for (auto cb: set_immediate)
-    {
-        cb();
-    }
-    set_immediate.clear();
+    handle_immediate_ops();
    return ret;
 }

+void osd_messenger_t::clear_immediate_ops(int peer_fd)
+{
+    size_t i = 0, j = 0;
+    while (i < set_immediate_ops.size())
+    {
+        if (set_immediate_ops[i]->peer_fd == peer_fd)
+        {
+            delete set_immediate_ops[i];
+        }
+        else
+        {
+            if (i != j)
+                set_immediate_ops[j] = set_immediate_ops[i];
+            j++;
+        }
+        i++;
+    }
+    set_immediate_ops.resize(j);
+}
+
+void osd_messenger_t::handle_immediate_ops()
+{
+    for (auto op: set_immediate_ops)
+    {
+        if (op->op_type == OSD_OP_IN)
+        {
+            exec_op(op);
+        }
+        else
+        {
+            // Copy lambda to be unaffected by `delete op`
+            std::function<void(osd_op_t*)>(op->callback)(op);
+        }
+    }
+    set_immediate_ops.clear();
+}
+
 bool osd_messenger_t::handle_read_buffer(osd_client_t *cl, void *curbuf, int remain)
 {
    // Compose operation(s) from the buffer
@@ -199,7 +234,7 @@ bool osd_messenger_t::handle_finished_read(osd_client_t *cl)
    {
        // Operation is ready
        cl->received_ops.push_back(cl->read_op);
-        set_immediate.push_back([this, op = cl->read_op]() { exec_op(op); });
+        set_immediate_ops.push_back(cl->read_op);
        cl->read_op = NULL;
        cl->read_state = 0;
    }
@@ -295,7 +330,7 @@ void osd_messenger_t::handle_op_hdr(osd_client_t *cl)
    {
        // Operation is ready
        cl->received_ops.push_back(cur_op);
-        set_immediate.push_back([this, cur_op]() { exec_op(cur_op); });
+        set_immediate_ops.push_back(cur_op);
        cl->read_op = NULL;
        cl->read_state = 0;
    }
@@ -416,9 +451,5 @@ void osd_messenger_t::handle_reply_ready(osd_op_t *op)
        (tv_end.tv_sec - op->tv_begin.tv_sec)*1000000 +
        (tv_end.tv_nsec - op->tv_begin.tv_nsec)/1000
    );
-    set_immediate.push_back([op]()
-    {
-        // Copy lambda to be unaffected by `delete op`
-        std::function<void(osd_op_t*)>(op->callback)(op);
-    });
+    set_immediate_ops.push_back(op);
 }
@@ -64,7 +64,7 @@ static void netlink_sock_alloc(struct netlink_ctx *ctx)
    if (nl_driver_id < 0)
    {
        nl_socket_free(sk);
-        fail("Couldn't resolve the nbd netlink family\n");
+        fail("Couldn't resolve the nbd netlink family: %s (code %d)\n", nl_geterror(nl_driver_id), nl_driver_id);
    }

    ctx->driver_id = nl_driver_id;
@@ -555,7 +555,12 @@ help:
        fcntl(sockfd[0], F_SETFL, fcntl(sockfd[0], F_GETFL, 0) | O_NONBLOCK);
        nbd_fd = sockfd[0];
        load_module();
+
        bool bg = cfg["foreground"].is_null();
+        if (cfg["logfile"].string_value() != "")
+        {
+            logfile = cfg["logfile"].string_value();
+        }

        if (netlink)
        {
@@ -579,6 +584,10 @@ help:
            if (!cfg["nbd_disconnect_on_close"].is_null())
                cflags |= NBD_CFLAG_DISCONNECT_ON_CLOSE;
 #endif
+            if (bg)
+            {
+                daemonize_fork();
+            }
            int err = netlink_configure(sockfd + 1, 1, devnum, device_size, 4096, flags, cflags, nbd_timeout, nbd_conn_timeout, NULL, revive);
            if (err < 0)
            {
@@ -588,6 +597,10 @@ help:
            }
            close(sockfd[1]);
            printf("/dev/nbd%d\n", err);
+            if (bg)
+            {
+                daemonize_reopen_stdio();
+            }
 #else
            fprintf(stderr, "netlink support is disabled in this build\n");
            exit(1);
@@ -631,14 +644,10 @@ help:
                    }
                }
            }
-        }
-        if (cfg["logfile"].string_value() != "")
-        {
-            logfile = cfg["logfile"].string_value();
-        }
-        if (bg)
-        {
-            daemonize();
+            if (bg)
+            {
+                daemonize();
+            }
        }
        // Initialize read state
        read_state = CL_READ_HDR;
@@ -716,13 +725,17 @@ help:
        }
    }

-    void daemonize()
+    void daemonize_fork()
    {
        if (fork())
            exit(0);
        setsid();
        if (fork())
            exit(0);
+    }
+
+    void daemonize_reopen_stdio()
+    {
        close(0);
        close(1);
        close(2);
@@ -733,6 +746,12 @@ help:
            fprintf(stderr, "Warning: Failed to chdir into /\n");
    }

+    void daemonize()
+    {
+        daemonize_fork();
+        daemonize_reopen_stdio();
+    }
+
    json11::Json::object list_mapped()
    {
        const char *self_filename = exe_name;
@@ -783,8 +802,9 @@ help:
                if (!strcmp(pid_filename, self_filename))
                {
                    json11::Json::object cfg = nbd_proxy::parse_args(argv.size(), argv.data());
-                    if (cfg["command"] == "map")
+                    if (cfg["command"] == "map" || cfg["command"] == "netlink-map")
                    {
+                        cfg["interface"] = (cfg["command"] == "netlink-map") ? "netlink" : "nbd";
                        cfg.erase("command");
                        cfg["pid"] = pid;
                        mapped["/dev/nbd"+std::to_string(dev_num)] = cfg;
@@ -16,7 +16,6 @@
 #include "qapi/error.h"
 #include "qapi/qmp/qdict.h"
 #include "qapi/qmp/qerror.h"
-#include "qemu/uri.h"
 #include "qemu/error-report.h"
 #include "qemu/module.h"
 #include "qemu/option.h"
@@ -1021,7 +1020,11 @@ static BlockDriver bdrv_vitastor = {
    // FIXME: Implement it along with per-inode statistics
    //.bdrv_get_allocated_file_size   = vitastor_get_allocated_file_size,

+#if QEMU_VERSION_MAJOR > 9 || QEMU_VERSION_MAJOR == 9 && QEMU_VERSION_MINOR > 0
+    .bdrv_open                      = vitastor_file_open,
+#else
    .bdrv_file_open                 = vitastor_file_open,
+#endif
    .bdrv_close                     = vitastor_close,

    // Option list for the create operation
@@ -6,7 +6,7 @@ includedir=${prefix}/@CMAKE_INSTALL_INCLUDEDIR@

 Name: Vitastor
 Description: Vitastor client library
-Version: 1.8.0
+Version: 1.10.0
 Libs: -L${libdir} -lvitastor_client
 Cflags: -I${includedir}

@@ -261,6 +261,7 @@ struct dd_out_info_t
            else
            {
                // ok
+                out_size = owatch->cfg.size;
                return true;
            }
            // Wait for sub-command
@@ -323,7 +324,7 @@ resume_2:
                cluster_op_t *sync_op = new cluster_op_t;
                sync_op->opcode = OSD_OP_SYNC;
                parent->waiting++;
-                sync_op->callback = [this, parent](cluster_op_t *sync_op)
+                sync_op->callback = [parent](cluster_op_t *sync_op)
                {
                    parent->waiting--;
                    delete sync_op;
@@ -369,6 +370,7 @@ struct cli_dd_t
 {
    cli_tool_t *parent;

+    std::vector<std::string> conv, iflag, oflag;
    dd_in_info_t iinfo;
    dd_out_info_t oinfo;

@@ -430,7 +432,7 @@ struct cli_dd_t
            if (read_op->retval < 0)
            {
                fprintf(
-                    stderr, "Failed to read bitmap for %lu bytes from image %s at offset %lu: %s (code %d)\n",
+                    stderr, "Failed to read bitmap for %ju bytes from image %s at offset %ju: %s (code %d)\n",
                    read_op->len, iinfo.iimg.c_str(), read_op->offset,
                    strerror(read_op->retval < 0 ? -read_op->retval : EIO), read_op->retval
                );
@@ -475,7 +477,7 @@ struct cli_dd_t
            if (read_op->retval != read_op->len)
            {
                fprintf(
-                    stderr, "Failed to read %lu bytes from image %s at offset %lu: %s (code %d)\n",
+                    stderr, "Failed to read %ju bytes from image %s at offset %ju: %s (code %d)\n",
                    read_op->len, iinfo.iimg.c_str(), read_op->offset,
                    strerror(read_op->retval < 0 ? -read_op->retval : EIO), read_op->retval
                );
@@ -537,7 +539,7 @@ struct cli_dd_t
                    in_eof = true;
            }
            ring_data_t *data = ((ring_data_t*)sqe->user_data);
-            data->iov = (iovec){ cur_read->buf + cur_read->len, cur_read->max - cur_read->len };
+            data->iov = (iovec){ (uint8_t*)cur_read->buf + cur_read->len, cur_read->max - cur_read->len };
            my_uring_prep_readv(sqe, iinfo.ifd, &data->iov, 1, iinfo.in_seekable ? iseek + cur_read->offset + cur_read->len : -1);
            in_waiting++;
            data->callback = [this, cur_read](ring_data_t *data)
@@ -546,7 +548,7 @@ struct cli_dd_t
                if (data->res < 0)
                {
                    fprintf(
-                        stderr, "Failed to read %lu bytes from %s at offset %lu: %s (code %d)\n",
+                        stderr, "Failed to read %ju bytes from %s at offset %ju: %s (code %d)\n",
                        data->iov.iov_len, iinfo.ifile == "" ? "stdin" : iinfo.ifile.c_str(), cur_read->offset,
                        strerror(-data->res), data->res
                    );
@@ -643,7 +645,7 @@ struct cli_dd_t
                if (write_op->retval != write_op->len)
                {
                    fprintf(
-                        stderr, "Failed to write %lu bytes to image %s at offset %lu: %s (code %d)\n",
+                        stderr, "Failed to write %ju bytes to image %s at offset %ju: %s (code %d)\n",
                        write_op->len, oinfo.oimg.c_str(), write_op->offset,
                        strerror(write_op->retval < 0 ? -write_op->retval : EIO), write_op->retval
                    );
@@ -670,7 +672,7 @@ struct cli_dd_t
                return false;
            }
            ring_data_t *data = ((ring_data_t*)sqe->user_data);
-            data->iov = (iovec){ .iov_base = cur_read->buf+cur_read->len, .iov_len = cur_read->max-cur_read->len };
+            data->iov = (iovec){ .iov_base = (uint8_t*)cur_read->buf+cur_read->len, .iov_len = cur_read->max-cur_read->len };
            my_uring_prep_writev(sqe, oinfo.ofd, &data->iov, 1, oinfo.out_seekable ? cur_read->offset+cur_read->len+oseek : -1);
            out_waiting++;
            data->callback = [this, cur_read](ring_data_t *data)
@@ -679,7 +681,7 @@ struct cli_dd_t
                if (data->res < 0)
                {
                    fprintf(
-                        stderr, "Failed to write %lu bytes to %s at offset %lu: %s (code %d)\n",
+                        stderr, "Failed to write %ju bytes to %s at offset %ju: %s (code %d)\n",
                        data->iov.iov_len, oinfo.ofile == "" ? "stdout" : oinfo.ofile.c_str(),
                        oinfo.out_seekable ? cur_read->offset+cur_read->len+oseek : 0,
                        strerror(-data->res), data->res
@@ -726,7 +728,7 @@ struct cli_dd_t
        {
            char buf[256];
            snprintf(
-                buf, sizeof(buf), "%lu bytes (%s) copied, %.1f s, %sB/s",
+                buf, sizeof(buf), "%ju bytes (%s) copied, %.1f s, %sB/s",
                written_size, format_size(written_size).c_str(), sec_total,
                format_size((uint64_t)(written_size/sec_total), true).c_str()
            );
@@ -748,7 +750,7 @@ struct cli_dd_t
        else
        {
            fprintf(
-                stderr, "\r%lu bytes (%s) copied, %.1f s, %sB/s, avg %sB/s\033[K",
+                stderr, "\r%ju bytes (%s) copied, %.1f s, %sB/s, avg %sB/s\033[K",
                written_size, format_size(written_size).c_str(), sec_total,
                format_size((uint64_t)(delta/sec_delta), true).c_str(),
                format_size((uint64_t)(written_size/sec_total), true).c_str()
@@ -766,6 +768,49 @@ struct cli_dd_t
            goto resume_3;
        else if (state == 4)
            goto resume_4;
+        for (int i = 0; i < conv.size(); i++)
+        {
+            if (conv[i] == "nofsync")
+                oinfo.end_fsync = false;
+            else if (conv[i] == "trunc")
+                oinfo.out_trunc = true;
+            else if (conv[i] == "nocreat")
+                oinfo.out_create = false;
+            else if (conv[i] == "noerror")
+                ignore_errors = true;
+            else if (conv[i] == "nosparse")
+                write_zero = true;
+            else
+            {
+                result = (cli_result_t){ .err = EINVAL, .text = "Unknown option conv="+conv[i] };
+                state = 100;
+                return;
+            }
+        }
+        for (int i = 0; i < iflag.size(); i++)
+        {
+            if (iflag[i] == "direct")
+                iinfo.in_direct = true;
+            else
+            {
+                result = (cli_result_t){ .err = EINVAL, .text = "Unknown option iflag="+iflag[i] };
+                state = 100;
+                return;
+            }
+        }
+        for (int i = 0; i < oflag.size(); i++)
+        {
+            if (oflag[i] == "direct")
+                oinfo.out_direct = true;
+            else if (oflag[i] == "append")
+                oinfo.out_append = true;
+            else
+            {
+                result = (cli_result_t){ .err = EINVAL, .text = "Unknown option oflag="+oflag[i] };
+                state = 100;
+                return;
+            }
+        }
        if ((oinfo.oimg != "" && oinfo.ofile != "") || (iinfo.iimg != "" && iinfo.ifile != ""))
        {
            result = (cli_result_t){ .err = EINVAL, .text = "Image and file can't be specified at the same time" };
@@ -838,6 +883,8 @@ resume_2:
        oinfo.end_fsync = oinfo.end_fsync && oinfo.out_seekable;
        read_offset = 0;
        read_end = iinfo.in_seekable ? iinfo.in_size-iseek : 0;
+        if (oinfo.out_size && (!read_end || read_end > oinfo.out_size-oseek))
+            read_end = oinfo.out_size-oseek;
        if (bytelimit && (!read_end || read_end > bytelimit))
            read_end = bytelimit;
        clock_gettime(CLOCK_REALTIME, &tv_begin);
@@ -908,6 +955,18 @@ static uint64_t parse_blocks(json11::Json v, uint64_t bs, uint64_t def)
    return res;
 }

+static std::vector<std::string> explode_json(const std::string & sep, json11::Json opt)
+{
+    if (opt.is_array())
+    {
+        std::vector<std::string> arr;
+        for (auto & item: opt.array_items())
+            arr.push_back(item.as_string());
+        return arr;
+    }
+    return explode(sep, opt.as_string(), true);
+}
+
 std::function<bool(cli_result_t &)> cli_tool_t::start_dd(json11::Json cfg)
 {
    auto dd = new cli_dd_t();
@@ -923,7 +982,7 @@ std::function<bool(cli_result_t &)> cli_tool_t::start_dd(json11::Json cfg)
    dd->oseek = parse_blocks(cfg["oseek"], dd->blocksize, 0);
    if (!dd->oseek)
        dd->oseek = parse_blocks(cfg["seek"], dd->blocksize, 0);
-    dd->iseek = parse_blocks(cfg["oseek"], dd->blocksize, 0);
+    dd->iseek = parse_blocks(cfg["iseek"], dd->blocksize, 0);
    if (!dd->iseek)
        dd->iseek = parse_blocks(cfg["skip"], dd->blocksize, 0);
    dd->iodepth = cfg["iodepth"].uint64_value();
@@ -935,25 +994,9 @@ std::function<bool(cli_result_t &)> cli_tool_t::start_dd(json11::Json cfg)
        progress = true;
    dd->iinfo.detect_size = cfg["size"].is_null();
    dd->oinfo.out_size = parse_size(cfg["size"].as_string());
-    std::vector<std::string> conv = explode(",", cfg["conv"].string_value(), true);
-    if (std::find(conv.begin(), conv.end(), "nofsync") != conv.end())
-        dd->oinfo.end_fsync = false;
-    if (std::find(conv.begin(), conv.end(), "trunc") != conv.end())
-        dd->oinfo.out_trunc = true;
-    if (std::find(conv.begin(), conv.end(), "nocreat") != conv.end())
-        dd->oinfo.out_create = false;
-    if (std::find(conv.begin(), conv.end(), "noerror") != conv.end())
-        dd->ignore_errors = true;
-    if (std::find(conv.begin(), conv.end(), "nosparse") != conv.end())
-        dd->write_zero = true;
-    conv = explode(",", cfg["iflag"].string_value(), true);
-    if (std::find(conv.begin(), conv.end(), "direct") != conv.end())
-        dd->iinfo.in_direct = true;
-    conv = explode(",", cfg["oflag"].string_value(), true);
-    if (std::find(conv.begin(), conv.end(), "direct") != conv.end())
-        dd->oinfo.out_direct = true;
-    if (std::find(conv.begin(), conv.end(), "append") != conv.end())
-        dd->oinfo.out_append = true;
+    dd->conv = explode_json(",", cfg["conv"]);
+    dd->iflag = explode_json(",", cfg["iflag"]);
+    dd->oflag = explode_json(",", cfg["oflag"]);
    return [dd](cli_result_t & result)
    {
        dd->loop();
@@ -5,6 +5,7 @@
 #include "cluster_client.h"
 #include "pg_states.h"
 #include "str_util.h"
+#include "json_util.h"

 struct cli_fix_t
 {
@@ -21,6 +21,3 @@ template<class T> void remove_duplicates(std::vector<T> & ret)
    }
    ret.resize(j+1);
 }
-
-// from http_client.cpp...
-bool json_is_false(const json11::Json & val);
@@ -4,6 +4,7 @@
 #include "cli.h"
 #include "cluster_client.h"
 #include "str_util.h"
+#include "json_util.h"
 #include "http_client.h"

 // Reweight OSD, change tags or set noout flag
@@ -156,6 +156,8 @@ resume_1:
                    for (auto & jtag: osd_cfg["tags"].array_items())
                        osd.tags.push_back(jtag.string_value());
                }
+                else if (osd_cfg["tags"].is_string())
+                    osd.tags.push_back(osd_cfg["tags"].string_value());
                osd.noout = osd_cfg["noout"].bool_value();
            }
            auto np_it = node_placement.find(std::to_string(osd.num));
@@ -214,7 +216,7 @@ resume_1:
                for (uint64_t osd_num: node.child_osds)
                {
                    auto & osd = placement_tree->osds.at(osd_num);
-                    fmt_items.push_back(json11::Json::object{
+                    auto json_osd = json11::Json::object{
                        { "type", "osd" },
                        { "name", osd.num },
                        { "parent", node.name },
@@ -228,7 +230,16 @@ resume_1:
                        { "bitmap", (uint64_t)osd.bitmap_granularity },
                        { "commit", osd.immediate_commit == IMMEDIATE_NONE ? "none" : (osd.immediate_commit == IMMEDIATE_ALL ? "all" : "small") },
                        { "op_stats", osd_stats[osd_num]["op_stats"] },
-                    });
+                    };
+                    if (osd_stats[osd_num]["slow_ops_primary"].uint64_value() > 0)
+                    {
+                        json_osd["slow_ops_primary"] = osd_stats[osd_num]["slow_ops_primary"];
+                    }
+                    if (osd_stats[osd_num]["slow_ops_secondary"].uint64_value() > 0)
+                    {
+                        json_osd["slow_ops_secondary"] = osd_stats[osd_num]["slow_ops_secondary"];
+                    }
+                    fmt_items.push_back(json_osd);
                }
            }
            result.data = fmt_items;
@@ -35,6 +35,7 @@ struct pool_creator_t
    uint64_t new_pools_mod_rev;
    json11::Json state_node_tree;
    json11::Json new_pools;
+    std::map<osd_num_t, json11::Json> osd_stats;

    bool is_done() { return state == 100; }

@@ -46,8 +47,6 @@ struct pool_creator_t
            goto resume_2;
        else if (state == 3)
            goto resume_3;
-        else if (state == 4)
-            goto resume_4;
        else if (state == 5)
            goto resume_5;
        else if (state == 6)
@@ -90,13 +89,19 @@ resume_1:
        // If not forced, check that we have enough osds for pg_size
        if (!force)
        {
-            // Get node_placement configuration from etcd
+            // Get node_placement configuration from etcd and OSD stats
            parent->etcd_txn(json11::Json::object {
                { "success", json11::Json::array {
                    json11::Json::object {
                        { "request_range", json11::Json::object {
                            { "key", base64_encode(parent->cli->st_cli.etcd_prefix+"/config/node_placement") },
-                        } }
+                        } },
+                    },
+                    json11::Json::object {
+                        { "request_range", json11::Json::object {
+                            { "key", base64_encode(parent->cli->st_cli.etcd_prefix+"/osd/stats/") },
+                            { "range_end", base64_encode(parent->cli->st_cli.etcd_prefix+"/osd/stats0") },
+                        } },
                    },
                } },
            });
@@ -112,10 +117,21 @@ resume_2:
                return;
            }

-            // Get state_node_tree based on node_placement and osd peer states
+            // Get state_node_tree based on node_placement and osd stats
            {
-                auto kv = parent->cli->st_cli.parse_etcd_kv(parent->etcd_result["responses"][0]["response_range"]["kvs"][0]);
-                state_node_tree = get_state_node_tree(kv.value.object_items());
+                auto node_placement_kv = parent->cli->st_cli.parse_etcd_kv(parent->etcd_result["responses"][0]["response_range"]["kvs"][0]);
+                timespec tv_now;
+                clock_gettime(CLOCK_REALTIME, &tv_now);
+                uint64_t osd_out_time = parent->cli->config["osd_out_time"].uint64_value();
+                if (!osd_out_time)
+                    osd_out_time = 600;
+                osd_stats.clear();
+                parent->iterate_kvs_1(parent->etcd_result["responses"][1]["response_range"]["kvs"], "/osd/stats/", [&](uint64_t cur_osd, json11::Json value)
+                {
+                    if ((uint64_t)value["time"].number_value()+osd_out_time >= tv_now.tv_sec)
+                        osd_stats[cur_osd] = value;
+                });
+                state_node_tree = get_state_node_tree(node_placement_kv.value.object_items(), osd_stats);
            }

            // Skip tag checks, if pool has none
@@ -158,42 +174,18 @@ resume_3:
                }
            }

-            // Get stats (for block_size, bitmap_granularity, ...) of osds in state_node_tree
-            {
-                json11::Json::array osd_stats;
-
-                for (auto osd_num: state_node_tree["osds"].array_items())
-                {
-                    osd_stats.push_back(json11::Json::object {
-                        { "request_range", json11::Json::object {
-                            { "key", base64_encode(parent->cli->st_cli.etcd_prefix+"/osd/stats/"+osd_num.as_string()) },
-                        } }
-                    });
-                }
-
-                parent->etcd_txn(json11::Json::object{ { "success", osd_stats } });
-            }
-
-            state = 4;
-resume_4:
-            if (parent->waiting > 0)
-                return;
-            if (parent->etcd_err.err)
-            {
-                result = parent->etcd_err;
-                state = 100;
-                return;
-            }
-
            // Filter osds from state_node_tree based on pool parameters and osd stats
            {
-                std::vector<json11::Json> osd_stats;
-                for (auto & ocr: parent->etcd_result["responses"].array_items())
+                std::vector<json11::Json> filtered_osd_stats;
+                for (auto & osd_num: state_node_tree["osds"].array_items())
                {
-                    auto kv = parent->cli->st_cli.parse_etcd_kv(ocr["response_range"]["kvs"][0]);
-                    osd_stats.push_back(kv.value);
+                    auto st_it = osd_stats.find(osd_num.uint64_value());
+                    if (st_it != osd_stats.end())
+                    {
+                        filtered_osd_stats.push_back(st_it->second);
+                    }
                }
-                guess_block_size(osd_stats);
+                guess_block_size(filtered_osd_stats);
                state_node_tree = filter_state_node_tree_by_stats(state_node_tree, osd_stats);
            }

@@ -201,8 +193,7 @@ resume_4:
            {
                auto failure_domain = cfg["failure_domain"].string_value() == ""
                    ? "host" : cfg["failure_domain"].string_value();
-                uint64_t max_pg_size = get_max_pg_size(state_node_tree["nodes"].object_items(),
-                    failure_domain, cfg["root_node"].string_value());
+                uint64_t max_pg_size = get_max_pg_size(state_node_tree, failure_domain, cfg["root_node"].string_value());

                if (cfg["pg_size"].uint64_value() > max_pg_size)
                {
@@ -358,56 +349,50 @@ resume_8:

    // Returns a JSON object of form {"nodes": {...}, "osds": [...]} that
    // contains: all nodes (osds, hosts, ...) based on node_placement config
-    // and current peer state, and a list of active peer osds.
-    json11::Json get_state_node_tree(json11::Json::object node_placement)
+    // and current osd stats.
+    json11::Json get_state_node_tree(json11::Json::object node_placement, std::map<osd_num_t, json11::Json> & osd_stats)
    {
-        // Erase non-peer osd nodes from node_placement
+        // Erase non-existing osd nodes from node_placement
        for (auto np_it = node_placement.begin(); np_it != node_placement.end();)
        {
            // Numeric nodes are osds
            osd_num_t osd_num = stoull_full(np_it->first);

-            // If node is osd and it is not in peer states, erase it
-            if (osd_num > 0 &&
-                parent->cli->st_cli.peer_states.find(osd_num) == parent->cli->st_cli.peer_states.end())
-            {
+            // If node is osd and its stats do not exist, erase it
+            if (osd_num > 0 && osd_stats.find(osd_num) == osd_stats.end())
                node_placement.erase(np_it++);
-            }
            else
                np_it++;
        }

-        // List of peer osds
-        std::vector<std::string> peer_osds;
+        // List of osds
+        std::vector<std::string> existing_osds;

-        // Record peer osds and add missing osds/hosts to np
-        for (auto & ps: parent->cli->st_cli.peer_states)
+        // Record osds and add missing osds/hosts to np
+        for (auto & ps: osd_stats)
        {
            std::string osd_num = std::to_string(ps.first);

-            // Record peer osd
-            peer_osds.push_back(osd_num);
+            // Record osd
+            existing_osds.push_back(osd_num);

-            // Add osd, if necessary
-            if (node_placement.find(osd_num) == node_placement.end())
+            // Add host if necessary
+            std::string osd_host = ps.second["host"].as_string();
+            if (node_placement.find(osd_host) == node_placement.end())
            {
-                std::string osd_host = ps.second["host"].as_string();
-
-                // Add host, if necessary
-                if (node_placement.find(osd_host) == node_placement.end())
-                {
-                    node_placement[osd_host] = json11::Json::object {
-                        { "level", "host" }
-                    };
-                }
-
-                node_placement[osd_num] = json11::Json::object {
-                    { "parent", osd_host }
+                node_placement[osd_host] = json11::Json::object {
+                    { "level", "host" }
                };
            }
+
+            // Add osd
+            node_placement[osd_num] = json11::Json::object {
+                { "parent", node_placement[osd_num]["parent"].is_null() ? osd_host : node_placement[osd_num]["parent"] },
+                { "level", "osd" },
+            };
        }

-        return json11::Json::object { { "osds", peer_osds }, { "nodes", node_placement } };
+        return json11::Json::object { { "osds", existing_osds }, { "nodes", node_placement } };
    }

    // Returns new state_node_tree based on given state_node_tree with osds
@@ -534,15 +519,13 @@ resume_8:
    // filtered out by stats parameters (block_size, bitmap_granularity) in
    // given osd_stats and current pool config.
    // Requires: state_node_tree["osds"] must match osd_stats 1-1
-    json11::Json filter_state_node_tree_by_stats(const json11::Json & state_node_tree, std::vector<json11::Json> & osd_stats)
+    json11::Json filter_state_node_tree_by_stats(const json11::Json & state_node_tree, std::map<osd_num_t, json11::Json> & osd_stats)
    {
-        auto & osds = state_node_tree["osds"].array_items();
-
        // Accepted state_node_tree nodes
        auto accepted_nodes = state_node_tree["nodes"].object_items();

        // List of accepted osds
-        std::vector<std::string> accepted_osds;
+        json11::Json::array accepted_osds;

        block_size = cfg["block_size"].uint64_value()
            ? cfg["block_size"].uint64_value()
@@ -554,21 +537,25 @@ resume_8:
            ? etcd_state_client_t::parse_immediate_commit(cfg["immediate_commit"].string_value(), IMMEDIATE_ALL)
            : parent->cli->st_cli.global_immediate_commit;

-        for (size_t i = 0; i < osd_stats.size(); i++)
+        for (auto osd_num_json: state_node_tree["osds"].array_items())
        {
-            auto & os = osd_stats[i];
-            // Get osd number
-            auto osd_num = osds[i].as_string();
+            auto osd_num = osd_num_json.uint64_value();
+            auto os_it = osd_stats.find(osd_num);
+            if (os_it == osd_stats.end())
+            {
+                continue;
+            }
+            auto & os = os_it->second;
            if (!os["data_block_size"].is_null() && os["data_block_size"] != block_size ||
                !os["bitmap_granularity"].is_null() && os["bitmap_granularity"] != bitmap_granularity ||
                !os["immediate_commit"].is_null() &&
                etcd_state_client_t::parse_immediate_commit(os["immediate_commit"].string_value(), IMMEDIATE_NONE) < immediate_commit)
            {
-                accepted_nodes.erase(osd_num);
+                accepted_nodes.erase(osd_num_json.as_string());
            }
            else
            {
-                accepted_osds.push_back(osd_num);
+                accepted_osds.push_back(osd_num_json);
            }
        }

@@ -576,87 +563,28 @@ resume_8:
    }

    // Returns maximum pg_size possible for given node_tree and failure_domain, starting at parent_node
-    uint64_t get_max_pg_size(json11::Json::object node_tree, const std::string & level, const std::string & parent_node)
+    uint64_t get_max_pg_size(json11::Json state_node_tree, const std::string & level, const std::string & root_node)
    {
-        uint64_t max_pg_sz = 0;
-
-        std::vector<std::string> nodes;
-
-        // Check if parent node is an osd (numeric)
-        if (parent_node != "" && stoull_full(parent_node))
+        std::set<std::string> level_seen;
+        for (auto & osd: state_node_tree["osds"].array_items())
        {
-            // Add it to node list if osd is in node tree
-            if (node_tree.find(parent_node) != node_tree.end())
-                nodes.push_back(parent_node);
-        }
-        // If parent node given, ...
-        else if (parent_node != "")
-        {
-            // ... look for children nodes of this parent
-            for (auto & sn: node_tree)
+            // find OSD parent at <level>, but stop at <root_node>
+            auto cur_id = osd.string_value();
+            auto cur = state_node_tree["nodes"][cur_id];
+            while (!cur.is_null())
            {
-                auto & props = sn.second.object_items();
-
-                auto parent_prop = props.find("parent");
-                if (parent_prop != props.end() && (parent_prop->second.as_string() == parent_node))
+                if (cur["level"] == level)
                {
-                    nodes.push_back(sn.first);
-
-                    // If we're not looking for all osds, we only need a single
-                    // child osd node
-                    if (level != "osd" && stoull_full(sn.first))
-                        break;
+                    level_seen.insert(cur_id);
+                    break;
                }
+                if (cur_id == root_node)
+                    break;
+                cur_id = cur["parent"].string_value();
+                cur = state_node_tree["nodes"][cur_id];
            }
        }
-        // No parent node given, and we're not looking for all osds
-        else if (level != "osd")
-        {
-            // ... look for all level nodes
-            for (auto & sn: node_tree)
-            {
-                auto & props = sn.second.object_items();
-
-                auto level_prop = props.find("level");
-                if (level_prop != props.end() && (level_prop->second.as_string() == level))
-                {
-                    nodes.push_back(sn.first);
-                }
-            }
-        }
-        // Otherwise, ...
-        else
-        {
-            // ... we're looking for osd nodes only
-            for (auto & sn: node_tree)
-            {
-                if (stoull_full(sn.first))
-                {
-                    nodes.push_back(sn.first);
-                }
-            }
-        }
-
-        // Process gathered nodes
-        for (auto & node: nodes)
-        {
-            // Check for osd node, return constant max size
-            if (stoull_full(node))
-            {
-                max_pg_sz += 1;
-            }
-            // Otherwise, ...
-            else
-            {
-                // ... exclude parent node from tree, and ...
-                node_tree.erase(parent_node);
-
-                // ... descend onto the resulting tree
-                max_pg_sz += get_max_pg_size(node_tree, level, node);
-            }
-        }
-
-        return max_pg_sz;
+        return level_seen.size();
    }

    json11::Json create_pool(const etcd_kv_t & kv)
@@ -4,6 +4,7 @@
 #include "cli.h"
 #include "cluster_client.h"
 #include "str_util.h"
+#include "json_util.h"
 #include "pg_states.h"
 #include "http_client.h"

@@ -133,6 +134,7 @@ resume_2:
        }
        int osd_count = 0, osd_up = 0;
        uint64_t total_raw = 0, free_raw = 0, free_down_raw = 0, down_raw = 0;
+        std::vector<uint64_t> slow_op_primary_osds, slow_op_secondary_osds;
        parent->iterate_kvs_1(osd_stats, "/osd/stats/", [&](uint64_t stat_osd_num, json11::Json value)
        {
            osd_count++;
@@ -140,18 +142,29 @@ resume_2:
            auto osd_free = value["free"].uint64_value();
            total_raw += osd_size;
            free_raw += osd_free;
-            if (!osd_free)
+            if (osd_size)
            {
-                osds_full++;
-            }
-            else if (osd_free < (uint64_t)(osd_size*(1-osd_nearfull_ratio)))
-            {
-                osds_nearfull++;
+                if (!osd_free)
+                {
+                    osds_full++;
+                }
+                else if (osd_free < (uint64_t)(osd_size*(1-osd_nearfull_ratio)))
+                {
+                    osds_nearfull++;
+                }
            }
            auto peer_it = parent->cli->st_cli.peer_states.find(stat_osd_num);
            if (peer_it != parent->cli->st_cli.peer_states.end())
            {
                osd_up++;
+                if (value["slow_ops_primary"].uint64_value() > 0)
+                {
+                    slow_op_primary_osds.push_back(stat_osd_num);
+                }
+                if (value["slow_ops_secondary"].uint64_value() > 0)
+                {
+                    slow_op_secondary_osds.push_back(stat_osd_num);
+                }
            }
            else
            {
@@ -215,6 +228,10 @@ resume_2:
                { "mon_master", mon_master },
                { "osd_up", osd_up },
                { "osd_count", osd_count },
+                { "osds_full", osds_full },
+                { "osds_nearfull", osds_nearfull },
+                { "osds_primary_slow_ops", slow_op_primary_osds },
+                { "osds_secondary_slow_ops", slow_op_secondary_osds },
                { "total_raw", total_raw },
                { "free_raw", free_raw },
                { "down_raw", down_raw },
@@ -299,6 +316,28 @@ resume_2:
            warning_str += "    "+std::to_string(osds_nearfull)+
                (osds_nearfull > 1 ? " osds are almost full\n" : " osd is almost full\n");
        }
+        if (slow_op_primary_osds.size() > 0)
+        {
+            warning_str += "    "+std::to_string(slow_op_primary_osds.size());
+            warning_str += (slow_op_primary_osds.size() > 1 ? " osds have" : " osd has");
+            warning_str += " slow client ops: ";
+            for (int i = 0; i < slow_op_primary_osds.size(); i++)
+            {
+                warning_str += (i > 0 ? ", " : "")+std::to_string(slow_op_primary_osds[i]);
+            }
+            warning_str += "\n";
+        }
+        if (slow_op_secondary_osds.size() > 0)
+        {
+            warning_str += "    "+std::to_string(slow_op_secondary_osds.size());
+            warning_str += (slow_op_secondary_osds.size() > 1 ? " osds have" : " osd has");
+            warning_str += " slow replication ops: ";
+            for (int i = 0; i < slow_op_secondary_osds.size(); i++)
+            {
+                warning_str += (i > 0 ? ", " : "")+std::to_string(slow_op_secondary_osds[i]);
+            }
+            warning_str += "\n";
+        }
        if (warning_str != "")
        {
            warning_str = "\n  warning:\n"+warning_str;
@@ -5,8 +5,9 @@ project(vitastor)
 # vitastor-disk
 add_executable(vitastor-disk
 	disk_tool.cpp disk_simple_offsets.cpp
-	disk_tool_journal.cpp disk_tool_meta.cpp disk_tool_prepare.cpp disk_tool_resize.cpp disk_tool_udev.cpp disk_tool_utils.cpp disk_tool_upgrade.cpp
-	../util/crc32c.c ../util/str_util.cpp ../../json11/json11.cpp ../util/rw_blocking.cpp ../util/allocator.cpp ../util/ringloop.cpp ../blockstore/blockstore_disk.cpp
+	disk_tool_journal.cpp disk_tool_meta.cpp disk_tool_prepare.cpp disk_tool_resize.cpp
+	disk_tool_resize_auto.cpp disk_tool_udev.cpp disk_tool_utils.cpp disk_tool_upgrade.cpp
+	../util/crc32c.c ../util/str_util.cpp ../util/json_util.cpp ../../json11/json11.cpp ../util/rw_blocking.cpp ../util/allocator.cpp ../util/ringloop.cpp ../blockstore/blockstore_disk.cpp
 )
 target_link_libraries(vitastor-disk
 	tcmalloc_minimal
@@ -27,12 +27,16 @@ static const char *help_text =
    "    --osd_per_disk <N>\n"
    "      Create <N> OSDs on each disk (default 1)\n"
    "    --hybrid\n"
-    "      Prepare hybrid (HDD+SSD) OSDs using provided devices. SSDs will be used for\n"
-    "      journals and metadata, HDDs will be used for data. Partitions for journals and\n"
-    "      metadata will be created automatically. Whether disks are SSD or HDD is decided\n"
-    "      by the `/sys/block/.../queue/rotational` flag. In hybrid mode, default object\n"
-    "      size is 1 MB instead of 128 KB, default journal size is 1 GB instead of 32 MB,\n"
-    "      and throttle_small_writes is enabled by default.\n"
+    "      Prepare hybrid (HDD+SSD, NVMe+SATA or etc) OSDs using provided devices. By default,\n"
+    "      any passed SSDs will be used for journals and metadata, HDDs will be used for data,\n"
+    "      but you can override this behaviour with --fast-devices option. Journal and metadata\n"
+    "      partitions will be created automatically. In the default mode, SSD and HDD disks\n"
+    "      are distinguished by the `/sys/block/.../queue/rotational` flag. When HDDs are used\n"
+    "      for data in hybrid mode, default block_size is 1 MB instead of 128 KB, default journal\n"
+    "      size is 1 GB instead of 32 MB, and throttle_small_writes is enabled by default.\n"
+    "    --fast-devices /dev/nvmeX,/dev/nvmeY\n"
+    "      In --hybrid mode, use these devices for journal and metadata instead of auto-detecting\n"
+    "      and extracting them from the main [devices...] list.\n"
    "    --disable_data_fsync auto\n"
    "      Disable data device cache and fsync (1/yes/true = on, default auto)\n"
    "    --disable_meta_fsync auto\n"
@@ -92,8 +96,22 @@ static const char *help_text =
    "  \n"
    "  Requires the `sfdisk` utility.\n"
    "\n"
-    "vitastor-disk resize <ALL_OSD_PARAMETERS> <NEW_LAYOUT> [--iodepth 32]\n"
-    "  Resize data area and/or rewrite/move journal and metadata\n"
+    "vitastor-disk resize <osd_num>|<osd_device> [OPTIONS]\n"
+    "  Resize data area and/or move journal and metadata:\n"
+    "    --move-journal TARGET    move journal to TARGET\n"
+    "    --move-meta TARGET       move metadata to TARGET\n"
+    "    --journal-size NEW_SIZE  resize journal to NEW_SIZE\n"
+    "    --data-size NEW_SIZE     resize data device to NEW_SIZE\n"
+    "    --dry-run                only show new layout, do not apply it\n"
+    "  \n"
+    "  NEW_SIZE may include k/m/g/t suffixes.\n"
+    "  TARGET may be one of:\n"
+    "    <partition>    move journal/metadata to an existing GPT partition\n"
+    "    <raw_device>   create a GPT partition on <raw_device> and move journal/metadata to it\n"
+    "    \"\"           (empty string) move journal/metadata back to the data device\n"
+    "\n"
+    "vitastor-disk raw-resize <ALL_OSD_PARAMETERS> <NEW_LAYOUT> [--iodepth 32]\n"
+    "  Resize data area and/or rewrite/move journal and metadata (manual format).\n"
    "  ALL_OSD_PARAMETERS must include all (at least all disk-related)\n"
    "  parameters from OSD command line (i.e. from systemd unit or superblock).\n"
    "  NEW_LAYOUT may include new disk layout parameters:\n"
@@ -143,8 +161,10 @@ static const char *help_text =
    "  For now, this only checks that device cache is in write-through mode if fsync is disabled.\n"
    "  Intended for use from startup scripts (i.e. from systemd units).\n"
    "\n"
+    "vitastor-disk dump-journal [OPTIONS] <osd_device>\n"
    "vitastor-disk dump-journal [OPTIONS] <journal_file> <journal_block_size> <offset> <size>\n"
-    "  Dump journal in human-readable or JSON (if --json is specified) format.\n"
+    "  Dump journal in text or JSON (if --json is specified) format.\n"
+    "  You can specify any OSD device (data, metadata or journal), or the layout manually.\n"
    "  Options:\n"
    "  --all             Scan the whole journal area for entries and dump them, even outdated ones\n"
    "  --json            Dump journal in JSON format\n"
@@ -152,16 +172,21 @@ static const char *help_text =
    "  --format data     Same as \"entries\", but also include small write data\n"
    "  --format blocks   Dump as an array of journal blocks each containing array of entries\n"
    "\n"
+    "vitastor-disk write-journal <osd_device>\n"
    "vitastor-disk write-journal <journal_file> <journal_block_size> <bitmap_size> <offset> <size>\n"
    "  Write journal from JSON taken from standard input in the same format as produced by\n"
    "  `dump-journal --json --format data`.\n"
+    "  You can specify any OSD device (data, metadata or journal), or the layout manually.\n"
    "\n"
+    "vitastor-disk dump-meta <osd_device>\n"
    "vitastor-disk dump-meta <meta_file> <meta_block_size> <offset> <size>\n"
    "  Dump metadata in JSON format.\n"
+    "  You can specify any OSD device (data, metadata or journal), or the layout manually.\n"
    "\n"
+    "vitastor-disk write-meta <osd_device>\n"
    "vitastor-disk write-meta <meta_file> <offset> <size>\n"
-    "  Write metadata from JSON taken from standard input in the same format as produced by\n"
-    "  `dump-meta`. Intended for debugging.\n"
+    "  Write metadata from JSON taken from standard input in the same format as produced by `dump-meta`.\n"
+    "  You can specify any OSD device (data, metadata or journal), or the layout manually.\n"
    "\n"
    "vitastor-disk simple-offsets <device>\n"
    "  Calculate offsets for old simple&stupid (no superblock) OSD deployment. Options:\n"
@@ -175,6 +200,7 @@ static const char *help_text =
    "    --device_size 0          Set device size\n"
    "    --format text            Result format: json, options, env, or text\n"
    "\n"
+    "Default I/O mode for commands involving disk I/O is O_DIRECT. If you don't want it, add --io cached.\n"
    "Use vitastor-disk --help <command> for command details or vitastor-disk --help --all for all details.\n"
 ;

@@ -199,6 +225,10 @@ int main(int argc, char *argv[])
        cmd.push_back((char*)"dump-journal");
        aliased = true;
    }
+    else if (!strcmp(exe_name, "vitastor-disk-test"))
+    {
+        self.test_mode = true;
+    }
    for (int i = 1; i < argc; i++)
    {
        if (!strcmp(argv[i], "--all"))
@@ -229,6 +259,10 @@ int main(int argc, char *argv[])
        {
            self.options["force"] = "1";
        }
+        else if (!strcmp(argv[i], "--dry-run") || !strcmp(argv[i], "--dry_run"))
+        {
+            self.options["dry_run"] = "1";
+        }
        else if (!strcmp(argv[i], "--allow-data-loss"))
        {
            self.options["allow_data_loss"] = "1";
@@ -236,7 +270,7 @@ int main(int argc, char *argv[])
        else if (argv[i][0] == '-' && argv[i][1] == '-' && i < argc-1)
        {
            char *key = argv[i]+2;
-            self.options[key] = argv[++i];
+            self.options[str_replace(key, "-", "_")] = argv[++i];
        }
        else
        {
@@ -249,29 +283,50 @@ int main(int argc, char *argv[])
    }
    if (!strcmp(cmd[0], "dump-journal"))
    {
-        if (cmd.size() < 5)
+        if (cmd.size() != 2 && cmd.size() < 5)
        {
            print_help(help_text, aliased ? "vitastor-dump-journal" : "vitastor-disk", cmd[0], false);
            return 1;
        }
        self.dsk.journal_device = cmd[1];
-        self.dsk.journal_block_size = strtoul(cmd[2], NULL, 10);
-        self.dsk.journal_offset = strtoull(cmd[3], NULL, 10);
-        self.dsk.journal_len = strtoull(cmd[4], NULL, 10);
+        if (cmd.size() > 2)
+        {
+            self.dsk.journal_block_size = strtoul(cmd[2], NULL, 10);
+            self.dsk.journal_offset = strtoull(cmd[3], NULL, 10);
+            self.dsk.journal_len = strtoull(cmd[4], NULL, 10);
+        }
+        else
+        {
+            // First argument is an OSD device - take metadata layout parameters from it
+            if (self.dump_load_check_superblock(self.dsk.journal_device))
+                return 1;
+        }
        return self.dump_journal();
    }
    else if (!strcmp(cmd[0], "write-journal"))
    {
-        if (cmd.size() < 6)
+        if (cmd.size() != 2 && cmd.size() < 6)
        {
            print_help(help_text, "vitastor-disk", cmd[0], false);
            return 1;
        }
        self.new_journal_device = cmd[1];
-        self.dsk.journal_block_size = strtoul(cmd[2], NULL, 10);
-        self.dsk.clean_entry_bitmap_size = strtoul(cmd[3], NULL, 10);
-        self.new_journal_offset = strtoull(cmd[4], NULL, 10);
-        self.new_journal_len = strtoull(cmd[5], NULL, 10);
+        if (cmd.size() > 2)
+        {
+            self.dsk.journal_block_size = strtoul(cmd[2], NULL, 10);
+            self.dsk.clean_entry_bitmap_size = strtoul(cmd[3], NULL, 10);
+            self.new_journal_offset = strtoull(cmd[4], NULL, 10);
+            self.new_journal_len = strtoull(cmd[5], NULL, 10);
+        }
+        else
+        {
+            // First argument is an OSD device - take metadata layout parameters from it
+            if (self.dump_load_check_superblock(self.new_journal_device))
+                return 1;
+            self.new_journal_device = self.dsk.journal_device;
+            self.new_journal_offset = self.dsk.journal_offset;
+            self.new_journal_len = self.dsk.journal_len;
+        }
        std::string json_err;
        json11::Json entries = json11::Json::parse(read_all_fd(0), json_err);
        if (json_err != "")
@@ -296,27 +351,48 @@ int main(int argc, char *argv[])
    }
    else if (!strcmp(cmd[0], "dump-meta"))
    {
-        if (cmd.size() < 5)
+        if (cmd.size() != 2 && cmd.size() < 5)
        {
            print_help(help_text, "vitastor-disk", cmd[0], false);
            return 1;
        }
        self.dsk.meta_device = cmd[1];
-        self.dsk.meta_block_size = strtoul(cmd[2], NULL, 10);
-        self.dsk.meta_offset = strtoull(cmd[3], NULL, 10);
-        self.dsk.meta_len = strtoull(cmd[4], NULL, 10);
+        if (cmd.size() > 2)
+        {
+            self.dsk.meta_block_size = strtoul(cmd[2], NULL, 10);
+            self.dsk.meta_offset = strtoull(cmd[3], NULL, 10);
+            self.dsk.meta_len = strtoull(cmd[4], NULL, 10);
+        }
+        else
+        {
+            // First argument is an OSD device - take metadata layout parameters from it
+            if (self.dump_load_check_superblock(self.dsk.meta_device))
+                return 1;
+        }
        return self.dump_meta();
    }
    else if (!strcmp(cmd[0], "write-meta"))
    {
-        if (cmd.size() < 4)
+        if (cmd.size() != 2 && cmd.size() < 4)
        {
            print_help(help_text, "vitastor-disk", cmd[0], false);
            return 1;
        }
        self.new_meta_device = cmd[1];
-        self.new_meta_offset = strtoull(cmd[2], NULL, 10);
-        self.new_meta_len = strtoull(cmd[3], NULL, 10);
+        if (cmd.size() > 2)
+        {
+            self.new_meta_offset = strtoull(cmd[2], NULL, 10);
+            self.new_meta_len = strtoull(cmd[3], NULL, 10);
+        }
+        else
+        {
+            // First argument is an OSD device - take metadata layout parameters from it
+            if (self.dump_load_check_superblock(self.new_meta_device))
+                return 1;
+            self.new_meta_device = self.dsk.meta_device;
+            self.new_meta_offset = self.dsk.meta_offset;
+            self.new_meta_len = self.dsk.meta_len;
+        }
        std::string json_err;
        json11::Json meta = json11::Json::parse(read_all_fd(0), json_err);
        if (json_err != "")
@@ -328,7 +404,16 @@ int main(int argc, char *argv[])
    }
    else if (!strcmp(cmd[0], "resize"))
    {
-        return self.resize_data();
+        if (cmd.size() != 2)
+        {
+            fprintf(stderr, "Exactly 1 OSD number or OSD device path argument is required\n");
+            return 1;
+        }
+        return self.resize_data(cmd[1]);
+    }
+    else if (!strcmp(cmd[0], "raw-resize"))
+    {
+        return self.raw_resize();
    }
    else if (!strcmp(cmd[0], "simple-offsets"))
    {
@@ -22,6 +22,7 @@
 #define VITASTOR_DISK_MAX_SB_SIZE 128*1024
 #define VITASTOR_PART_TYPE "e7009fac-a5a1-4d72-af72-53de13059903"
 #define DEFAULT_HYBRID_JOURNAL "1G"
+#define DEFAULT_HYBRID_SSD_JOURNAL "128M"

 struct resizer_data_moving_t;

@@ -40,6 +41,7 @@ struct disk_tool_t
    /**** Parameters ****/

    std::map<std::string, std::string> options;
+    bool test_mode = false;
    bool all, json, now;
    bool dump_with_blocks, dump_with_data;
    blockstore_disk_t dsk;
@@ -93,10 +95,16 @@ struct disk_tool_t
    void dump_meta_header(blockstore_meta_header_v2_t *hdr);
    void dump_meta_entry(uint64_t block_num, clean_disk_entry *entry, uint8_t *bitmap);

+    int dump_load_check_superblock(const std::string & device);
+
    int write_json_journal(json11::Json entries);
    int write_json_meta(json11::Json meta);

-    int resize_data();
+    int resize_data(std::string device);
+    int resize_parse_move_journal(std::map<std::string, std::string> & move_options, bool dry_run);
+    int resize_parse_move_meta(std::map<std::string, std::string> & move_options, bool dry_run);
+
+    int raw_resize();
    int resize_parse_params();
    void resize_init(blockstore_meta_header_v2_t *hdr);
    int resize_remap_blocks();
@@ -114,11 +122,14 @@ struct disk_tool_t
    int systemd_start_stop_osds(const std::vector<std::string> & cmd, const std::vector<std::string> & devices);
    int pre_exec_osd(std::string device);
    int purge_devices(const std::vector<std::string> & devices);
+    int clear_osd_superblock(const std::string & dev);

    json11::Json read_osd_superblock(std::string device, bool expect_exist = true, bool ignore_nonref = false);
    uint32_t write_osd_superblock(std::string device, json11::Json params);

    int prepare_one(std::map<std::string, std::string> options, int is_hdd = -1);
+    int check_existing_partition(std::string & dev_by_uuid);
+    int fix_partition_type(std::string & dev_by_uuid);
    int prepare(std::vector<std::string> devices);
    std::vector<vitastor_dev_info_t> collect_devices(const std::vector<std::string> & devices);
    json11::Json add_partitions(vitastor_dev_info_t & devinfo, std::vector<std::string> sizes);
@@ -133,13 +144,13 @@ void disk_tool_simple_offsets(json11::Json cfg, bool json_output);
 uint64_t sscanf_json(const char *fmt, const json11::Json & str);
 void fromhexstr(const std::string & from, int bytes, uint8_t *to);
 int disable_cache(std::string dev);
+uint64_t get_device_size(const std::string & dev, bool should_exist = false);
 std::string get_parent_device(std::string dev);
-bool json_is_true(const json11::Json & val);
 int shell_exec(const std::vector<std::string> & cmd, const std::string & in, std::string *out, std::string *err);
 int write_zero(int fd, uint64_t offset, uint64_t size);
 json11::Json read_parttable(std::string dev);
 uint64_t dev_size_from_parttable(json11::Json pt);
 uint64_t free_from_parttable(json11::Json pt);
-int fix_partition_type(std::string dev_by_uuid);
+int fix_partition_type_uuid(std::string & dev_by_uuid, const std::string & type_uuid);
 std::string csum_type_str(uint32_t data_csum_type);
 uint32_t csum_type_from_str(std::string data_csum_type);
@@ -18,7 +18,7 @@ int disk_tool_t::dump_journal()
        printf("[\n");
    if (all)
    {
-        dsk.journal_fd = open(dsk.journal_device.c_str(), O_DIRECT|O_RDONLY);
+        dsk.journal_fd = open(dsk.journal_device.c_str(), (options["io"] == "cached" ? 0 : O_DIRECT) | O_RDONLY);
        if (dsk.journal_fd < 0)
        {
            fprintf(stderr, "Failed to open journal device %s: %s\n", dsk.journal_device.c_str(), strerror(errno));
@@ -121,7 +121,7 @@ int disk_tool_t::dump_journal()

 int disk_tool_t::process_journal(std::function<int(void*)> block_fn)
 {
-    dsk.journal_fd = open(dsk.journal_device.c_str(), O_DIRECT|O_RDONLY);
+    dsk.journal_fd = open(dsk.journal_device.c_str(), (options["io"] == "cached" ? 0 : O_DIRECT) | O_RDONLY);
    if (dsk.journal_fd < 0)
    {
        fprintf(stderr, "Failed to open journal device %s: %s\n", dsk.journal_device.c_str(), strerror(errno));
@@ -517,6 +517,12 @@ int disk_tool_t::write_json_journal(json11::Json entries)
            uint32_t data_csum_size = !dsk.data_csum_type ? 0 : ne->small_write.len/dsk.csum_block_size*(dsk.data_csum_type & 0xFF);
            fromhexstr(rec["bitmap"].string_value(), dsk.clean_entry_bitmap_size, ((uint8_t*)ne) + sizeof(journal_entry_small_write) + data_csum_size);
            fromhexstr(rec["data"].string_value(), ne->small_write.len, new_journal_data);
+            if (ne->small_write.len > 0 && !rec["data"].is_string())
+            {
+                fprintf(stderr, "Error: entry data is missing, please generate the dump with --json --format data\n");
+                free(new_journal_buf);
+                return 1;
+            }
            if (dsk.data_csum_type)
                fromhexstr(rec["block_csums"].string_value(), data_csum_size, ((uint8_t*)ne) + sizeof(journal_entry_small_write));
            if (rec["data"].is_string())
@@ -4,6 +4,7 @@
 #include "disk_tool.h"
 #include "rw_blocking.h"
 #include "osd_id.h"
+#include "json_util.h"

 int disk_tool_t::process_meta(std::function<void(blockstore_meta_header_v2_t *)> hdr_fn,
    std::function<void(uint64_t, clean_disk_entry*, uint8_t*)> record_fn)
@@ -13,7 +14,7 @@ int disk_tool_t::process_meta(std::function<void(blockstore_meta_header_v2_t *)>
        fprintf(stderr, "Invalid metadata block size: is not a multiple of %d\n", DIRECT_IO_ALIGNMENT);
        return 1;
    }
-    dsk.meta_fd = open(dsk.meta_device.c_str(), O_DIRECT|O_RDONLY);
+    dsk.meta_fd = open(dsk.meta_device.c_str(), (options["io"] == "cached" ? 0 : O_DIRECT) | O_RDONLY);
    if (dsk.meta_fd < 0)
    {
        fprintf(stderr, "Failed to open metadata device %s: %s\n", dsk.meta_device.c_str(), strerror(errno));
@@ -149,6 +150,31 @@ int disk_tool_t::process_meta(std::function<void(blockstore_meta_header_v2_t *)>
    return 0;
 }

+int disk_tool_t::dump_load_check_superblock(const std::string & device)
+{
+    json11::Json sb = read_osd_superblock(device, true, false);
+    if (sb.is_null())
+        return 1;
+    try
+    {
+        auto cfg = json_to_string_map(sb["params"].object_items());
+        dsk.parse_config(cfg);
+        dsk.data_io = dsk.meta_io = dsk.journal_io = "cached";
+        dsk.open_data();
+        dsk.open_meta();
+        dsk.open_journal();
+        dsk.calc_lengths(true);
+    }
+    catch (std::exception & e)
+    {
+        dsk.close_all();
+        fprintf(stderr, "%s\n", e.what());
+        return 1;
+    }
+    dsk.close_all();
+    return 0;
+}
+
 int disk_tool_t::dump_meta()
 {
    int r = process_meta(
@@ -176,7 +202,7 @@ void disk_tool_t::dump_meta_header(blockstore_meta_header_v2_t *hdr)
        {
            printf(
                "{\"version\":\"0.9\",\"meta_block_size\":%u,\"data_block_size\":%u,\"bitmap_granularity\":%u,"
-                "\"data_csum_type\":%s,\"csum_block_size\":%u,\"entries\":[\n",
+                "\"data_csum_type\":\"%s\",\"csum_block_size\":%u,\"entries\":[\n",
                hdr->meta_block_size, hdr->data_block_size, hdr->bitmap_granularity,
                csum_type_str(hdr->data_csum_type).c_str(), hdr->csum_block_size
            );
@@ -243,12 +269,16 @@ int disk_tool_t::write_json_meta(json11::Json meta)
        ? meta["data_block_size"].uint64_value() : 131072;
    new_hdr->bitmap_granularity = meta["bitmap_granularity"].uint64_value()
        ? meta["bitmap_granularity"].uint64_value() : 4096;
-    new_hdr->data_csum_type = meta["data_csum_type"].is_number()
-        ? meta["data_csum_type"].uint64_value()
-        : (meta["data_csum_type"].string_value() == "crc32c"
-            ? BLOCKSTORE_CSUM_CRC32C
-            : BLOCKSTORE_CSUM_NONE);
-    new_hdr->csum_block_size = meta["csum_block_size"].uint64_value();
+    if (new_hdr->version >= BLOCKSTORE_META_FORMAT_V2)
+    {
+        new_hdr->data_csum_type = meta["data_csum_type"].is_number()
+            ? meta["data_csum_type"].uint64_value()
+            : (meta["data_csum_type"].string_value() == "crc32c"
+                ? BLOCKSTORE_CSUM_CRC32C
+                : BLOCKSTORE_CSUM_NONE);
+        new_hdr->csum_block_size = meta["csum_block_size"].uint64_value();
+        new_hdr->header_csum = crc32c(0, new_hdr, sizeof(*new_hdr));
+    }
    uint32_t new_clean_entry_header_size = (new_hdr->version == BLOCKSTORE_META_FORMAT_V1
        ? sizeof(clean_disk_entry) : sizeof(clean_disk_entry) + 4 /*entry_csum*/);
    new_clean_entry_bitmap_size = (new_hdr->data_block_size / new_hdr->bitmap_granularity + 7) / 8;
@@ -285,8 +315,7 @@ int disk_tool_t::write_json_meta(json11::Json meta)
                fromhexstr(e["data_csum"].string_value(), new_data_csum_size,
                    ((uint8_t*)new_entry) + sizeof(clean_disk_entry) + 2*new_clean_entry_bitmap_size);
            }
-            uint32_t *new_entry_csum = (uint32_t*)(((uint8_t*)new_entry) + sizeof(clean_disk_entry) +
-                2*new_clean_entry_bitmap_size + new_data_csum_size);
+            uint32_t *new_entry_csum = (uint32_t*)(((uint8_t*)new_entry) + new_clean_entry_size - 4);
            *new_entry_csum = crc32c(0, new_entry, new_clean_entry_size - 4);
        }
    }
--- a/Show More
+++ b/Show More
Author	SHA1	Message	Date
Vitaliy Filippov	7fbb04fdfa	Release 1.10.0 New features: - Implement basic VitastorFS support in [CSI](https://vitastor.io/docs/installation/kubernetes.html) - Implement [NFS RDMA](https://vitastor.io/docs/usage/nfs.html#rdma) support - Pause pool rebalance when monitor detects that it can lead to any OSD becoming full ([osd_backfillfull_ratio](https://vitastor.io/docs/config/monitor.html#osd_backfillfull_ratio)) - Auto-select correct [RDMA device and GID](https://vitastor.io/docs/config/network.html#rdma_device) based on osd_network and RoCEv2 priority - Report slow ops in OSD stats in etcd and show them in vitastor-cli status Bug fixes: - Fix possibly incorrect linked list deserialization in NFS - Fix possible crash in vitastor-nfs --block READDIR operation - Map netlink after forking to show correct PID in vitastor-nbd ls - Simplify and fix create-pool OSD count checks for the case of hosts split into sub-nodes - Make monitor print "Waiting to become master" just once, not every 5s - Take out_size from oimg if not specified in vitastor-cli dd - Do not report OSDs with empty statistics as "full" in status - Trigger double autosync when switching PG state to prevent leaving garbage in non-immediate_commit clusters - Fix a lack of connection timeout for etcd websockets in OSD leading to slower etcd failover (~70s instead of ~10s) - Fix a rare OSD crash during client disconnect - Fix PGs sometimes sticking until OSD restart in the "has_unclean" state with EC pools - Fix metadata partition zeroing in vitastor-disk prepare - Add patches for qemu 9.1 and pve-qemu 9.0 and 9.1 - Fix libvirt 8 patch	2024-12-19 15:49:19 +03:00
Vitaliy Filippov	63b85b6bfb	Fix clang warnings/errors	2024-12-19 15:30:31 +03:00
Vitaliy Filippov	2f5959e3fa	Add pve-qemu 9.1 patch	2024-12-19 14:05:12 +03:00
Vitaliy Filippov	a4a286ed95	Document NFS-RDMA	2024-12-19 14:05:12 +03:00
Vitaliy Filippov	b8009bad5e	Add librdmacm-dev to build dockerfile	2024-12-19 14:05:12 +03:00
Vitaliy Filippov	9be3d27dc9	Document VitastorFS-based CSI	2024-12-19 13:06:47 +03:00
Vitaliy Filippov	a19d2066c2	Document osd_backfillfull_ratio	2024-12-19 02:15:02 +03:00
Vitaliy Filippov	2a8780b4b5	Add a note about slow ops	2024-12-19 02:02:37 +03:00
Vitaliy Filippov	109f51a015	Implement basic VitastorFS support in CSI	2024-12-17 02:26:23 +03:00
Vitaliy Filippov	8a86c123c3	Allow to auto-select and print the port	2024-12-14 16:55:13 +03:00
Vitaliy Filippov	b856524e0c	Workaround for Linux bug: return post_op_attr for NFS-RDMA READ3 Linux NFS RDMA transport has a stupid bug - when the reply doesn't contain post_op_attr, the data gets offsetted by 84 bytes (size of attributes) and first 84 bytes are filled with probably random data.	2024-12-11 21:09:36 +03:00
Vitaliy Filippov	ae3ca7451f	Use per-connection RDMA device contexts	2024-12-11 21:09:36 +03:00
Vitaliy Filippov	1dbbb0c3f8	Implement NFS RDMA support	2024-12-11 21:09:36 +03:00
Vitaliy Filippov	64db31ec10	Fix slow op warning format	2024-12-11 21:09:36 +03:00
Vitaliy Filippov	76470686b3	Fix possibly incorrect linked list deserialization in NFS	2024-12-08 02:54:13 +03:00
Vitaliy Filippov	652ca631bb	Fix possible crash in nfs_block readdir	2024-12-01 18:04:49 +03:00
Vitaliy Filippov	2105f4b654	Add lost netlink daemonize	2024-11-27 17:13:30 +03:00
Vitaliy Filippov	0d01573da3	Fix typos	2024-11-26 14:31:47 +03:00
Vitaliy Filippov	d84b84f58d	Fix new backfillfull feature, add more logs	2024-11-23 01:08:13 +03:00
Vitaliy Filippov	8cfe705d7a	Map netlink after forking to show correct PID in vitastor-nbd ls	2024-11-23 00:46:44 +03:00
Vitaliy Filippov	66c9271cbd	Radically simplify create-pool pg_size check	2024-11-22 01:44:14 +03:00
Vitaliy Filippov	7b37ba921d	Pause pool rebalance when monitor detects that it can lead to any OSD becoming full	2024-11-22 01:01:07 +03:00
Vitaliy Filippov	262c581400	Fix create-pool for the case of hosts split into sub-nodes	2024-11-22 01:01:07 +03:00
Vitaliy Filippov	ad3b6b7267	Add a note about GID and RDMA device auto-selection	2024-11-21 23:54:05 +03:00
Vitaliy Filippov	1f6a061283	Move ibv_query_gid under #ifdef to only build it with libibverbs 32+	2024-11-21 23:47:57 +03:00
Vitaliy Filippov	fc4d97da10	Print "Waiting to become master" just once	2024-11-21 00:55:22 +03:00
Vitaliy Filippov	c7a4ce7341	Take out_size from dd oimg if not specified	2024-11-19 02:13:34 +03:00
Vitaliy Filippov	ddea31d86d	Auto-select first RDMA device only if RoCE is not found, add rocev2->rocev1->ib priority	2024-11-19 01:54:00 +03:00
Vitaliy Filippov	156d005412	Add serialize_overlap to test_heal	2024-11-17 01:26:35 +03:00
Vitaliy Filippov	7e076c7049	Do not report OSDs with empty statistics as full	2024-11-16 23:36:16 +03:00
Vitaliy Filippov	7de38250ad	Auto-select RDMA device based on osd_network	2024-11-16 18:38:57 +03:00
Vitaliy Filippov	9c59d30e83	Report slow ops in OSD stats in etcd and show them in vitastor-cli status	2024-11-16 15:11:16 +03:00
Vitaliy Filippov	5db02cdf6e	Add pve-qemu 9.0 patch	2024-11-16 11:20:47 +03:00
Vitaliy Filippov	8202ee9d74	Trigger double autosync when switching PG state to prevent leaving garbage in non-immediate_commit clusters	2024-11-15 01:26:36 +03:00
Vitaliy Filippov	5864bd067c	Add missing connection timeout for etcd websockets in OSD	2024-11-12 02:28:07 +03:00
Vitaliy Filippov	c312557ace	Do not execute remaining operations if the client is stopped during read	2024-11-10 16:44:13 +03:00
Vitaliy Filippov	5ce20116d8	Postpone trigger_nearest to prevent timer callbacks called from setTimer/clearTimer	2024-11-10 15:51:16 +03:00
Vitaliy Filippov	be66791e59	Add another note about 1.8 upgrade	2024-11-09 00:57:58 +03:00
Vitaliy Filippov	141cec2383	Add missing refcounting for flush_batch errors	2024-11-09 00:46:38 +03:00
Vitaliy Filippov	1ce4b1b417	Fix stop condition in osd_flush Could probably lead to PGs hung in peering states on OSD restart in EC pools, fixable by primary OSD restart	2024-11-08 00:30:40 +03:00
Vitaliy Filippov	ebf24bac9a	Fix partition zeroing during prepare Previously it zeroed area beginning with 0 instead of actual metadata offset which was leading to non-zeroed metadata when the disk is very small	2024-11-08 00:14:37 +03:00
Vitaliy Filippov	edd9051f81	Fix arch.en toc	2024-11-08 00:14:18 +03:00
Vitaliy Filippov	662ca86dc0	Fix libvirt 8 patch	2024-11-07 12:21:32 +03:00
Vitaliy Filippov	a1ca573168	Support QEMU 9.1	2024-11-07 12:21:13 +03:00
Vitaliy Filippov	f69f801ffb	Release 1.9.3 - Support custom hybrid OSD creation (`vitastor-disk prepare --hybrid --fast-devices /dev/xxx,/dev/yyy`) - Auto-change partition paths to /dev/disk/by-partuuid/ in `vitastor-disk prepare` - Allow to select cached I/O in vitastor-disk commands - Fix multiple bugs in vitastor-disk resize & add tests for them - Fix vitastor-disk write-meta/write-journal in superblock-based mode writing it to an incorrect device - Fix vitastor-disk prepare sometimes again not seeing new partitions - Cleanup PG history and stats of deleted pools - Fix "is already mounted" checks in CSI	2024-11-07 01:28:31 +03:00
Vitaliy Filippov	af92cbdfcc	Dynamic device size in test	2024-11-06 14:16:58 +03:00
Vitaliy Filippov	a775db10cc	Also allow cached I/O in dsk.open_*() in disk_tool	2024-11-06 13:52:25 +03:00
Vitaliy Filippov	eafce26049	Add resize and resize-auto tests	2024-11-06 13:30:51 +03:00
Vitaliy Filippov	625c74294f	Support direct I/O	2024-11-06 13:30:12 +03:00
Vitaliy Filippov	ef8c21ad6f	Change %lu to %ju	2024-11-06 02:58:51 +03:00
Vitaliy Filippov	2bb8e8999e	Do not check length in "data alignment mismatch"	2024-11-06 02:58:26 +03:00
Vitaliy Filippov	c2e7c28672	Fix calc_lengths data size recalc during auto-resize	2024-11-06 02:27:17 +03:00
Vitaliy Filippov	bd22beefb5	Auto-extend new_data_len if new_data_offset is changed too	2024-11-06 02:13:30 +03:00
Vitaliy Filippov	e7038ab99c	Auto-change partition paths to /dev/disk/by-partuuid/	2024-11-06 01:04:05 +03:00
Vitaliy Filippov	b6f75ebcfd	Add missing I/O path description in english	2024-11-06 00:43:17 +03:00
Vitaliy Filippov	9def199981	Auto-reduce new_data_len in resize	2024-11-05 02:57:11 +03:00
Vitaliy Filippov	c72e8e649e	Support test mode for vitastor-disk	2024-11-05 02:43:55 +03:00
Vitaliy Filippov	8bdb3e8786	Write meta/journal to correct device when used in superblock mode	2024-11-05 02:43:55 +03:00
Vitaliy Filippov	a87e236c70	Fix resize --data-size, particularly when expanding the device	2024-11-04 18:55:03 +03:00
Vitaliy Filippov	16f67cf6f1	Fix missing metadata checksums after resize	2024-11-04 18:36:35 +03:00
Vitaliy Filippov	56de4a520d	Support custom hybrid OSD creation (--hybrid --fast-devices /dev/xxx,/dev/yyy)	2024-11-04 17:52:29 +03:00
Vitaliy Filippov	adca162278	Note that osd_per_disk is also incompatible	2024-11-04 15:20:01 +03:00
Vitaliy Filippov	490b314d72	Rework & fix new partition waiting code	2024-11-04 15:16:30 +03:00
Vitaliy Filippov	9f52074e1e	Delete PG history and stats of deleted pools	2024-11-01 02:38:31 +03:00
Vitaliy Filippov	2b3e877546	Add notes about vitastor-disk in disable_data_fsync	2024-11-01 02:38:18 +03:00
Vitaliy FilippovandGitHub	01d55e5420	Merge pull request #64 from 0x00ace/fio_version_fix use fio 3.35-1 for AlmaLinux 9	2024-10-31 11:55:40 +03:00
Vitaliy Filippov	f5aa5cfdfe	Fix "is already mounted" checks in CSI	2024-10-26 14:06:21 +03:00
Vitaliy Filippov	2826bb9e7e	Add more logging to CSI	2024-10-24 02:07:55 +03:00
Vitaliy Filippov	30d1ad0f66	Add Intel D5-P4320	2024-10-22 23:22:48 +03:00
Vitaliy Filippov	79719e44ac	Release 1.9.2 New features: - Support resizing normal vitastor-disk partitions and moving journal/metadata: [vitastor-disk resize](https://vitastor.io/docs/usage/disk.html#resize) - Support simple forms of vitastor-disk {dump,write}-{meta,journal} for OSD partitions Bug fixes: - Fix block RWX volumes broken after introducing stage/unstage support - Do not allow to create non-block RWX volumes in CSI - Fix vitastor-disk prepare not seeing the newly created partition in rare cases - Fix non-array tags not showing up in ls-osd/osd-tree - Make OpenNebula oned.conf patching during installation smarter - Fix iseek option in vitastor-cli dd not working - Validate conv=, iflag=, oflag= options in vitastor-cli dd - Fix vitastor-disk write-meta not writing header checksum to the disk - Fix JSON format in vitastor-disk dump-meta - Fix read_chain_bitmap not working for snapshot in another pool - Fix a possible OSD crash during parallel read & write to an image with snapshots - Several followups to the READ_CHAIN_BITMAP fix: avoid data reads, fix possible overflow in is_zero(), fix bitmap size	2024-10-20 01:49:13 +03:00
Vitaliy Filippov	f5626655df	Add new disk command docs	2024-10-20 01:47:46 +03:00
Vitaliy Filippov	7e2dde2702	Fix block RWX volumes broken after introducing stage/unstage support	2024-10-19 11:56:56 +03:00
Vitaliy Filippov	3b0ab317cf	Validate non-block RWX in CSI	2024-10-18 01:55:38 +03:00
Vitaliy Filippov	18eb99c494	Implement resizing partitions created with vitastor-disk	2024-10-18 01:55:19 +03:00
Vitaliy Filippov	4e8a1a8895	Run partprobe in add_partition() if /dev/disk/by-partuuid symlink is not present	2024-10-12 18:07:53 +03:00
Vitaliy Filippov	d27a8bdabc	Make get_parent_device return full path	2024-10-12 13:44:52 +03:00
Vitaliy Filippov	ebd616e42f	Extract clear_osd_superblock()	2024-10-12 13:44:52 +03:00
Vitaliy Filippov	b18d296e01	Extract check_existing_partition(), get_device_size()	2024-10-12 13:44:52 +03:00
Vitaliy Filippov	a03508320e	Move json_is_true/json_is_false to json_util.cpp	2024-10-12 00:40:39 +03:00
Vitaliy Filippov	c9ccc790ec	Fix non-array tags not showing up in ls-osd/osd-tree	2024-10-11 18:33:35 +03:00
Vitaliy Filippov	db2d9c5b3d	Fix tables in NFS doc	2024-10-08 00:20:10 +03:00
Vitaliy Filippov	09f15f44c9	Fix Toshiba MG and VDUSE Debian kernel note in docs	2024-10-08 00:17:14 +03:00
Vitaliy Filippov	c5a58c2e81	Support reading parameters automatically from the superblock in vitastor-disk {dump,write}-{meta,journal}	2024-10-07 02:21:58 +03:00
Vitaliy Filippov	30e7c2ad1e	Add custom OpenNebula oned.conf patcher (it uses a SHITTY configuration file format)	2024-10-06 13:46:05 +03:00
Vitaliy Filippov	2e76ceabbe	Fix iseek option in vitastor-cli dd	2024-10-05 18:25:38 +03:00
Vitaliy Filippov	3df088c207	Validate conv=, iflag=, oflag= options in vitastor-cli dd	2024-10-05 18:02:36 +03:00
Vitaliy Filippov	d882a19eab	Fix vitastor-disk write-meta not writing header checksum to the disk...	2024-10-05 17:32:55 +03:00
Vitaliy Filippov	702be3da7a	Fix JSON format in vitastor-disk dump-meta	2024-10-05 16:08:34 +03:00
Vitaliy Filippov	99533e1c2f	Fix .yml links	2024-10-02 00:38:07 +03:00
Vitaliy Filippov	a6cceb43bf	Fix read_chain_bitmap not working for snapshot in another pool	2024-10-02 00:24:48 +03:00
Vitaliy Filippov	745d89459a	Fix link, add title	2024-09-29 22:05:56 +03:00
Vitaliy Filippov	48f023292d	Fix extra data reads on read_chain	2024-09-21 17:05:42 +03:00
Vitaliy Filippov	b58bf3ada5	Fix possible OSD crash during parallel read & write to an image with snapshots OSDs could crash with the following "assertion failed" message (crash didn't affect data and was caused by OSD thinking upper blocks are full while they weren't). Reproduction without introducing artificial delays is hard because you have to force OSD to read an object with enqueued but not handled write which fills previously non-full bitmap. O_o. ``` vitastor-osd: ./src/osd/osd_primary_chain.cpp:613: void osd_t::send_chained_read_results(pg_t&, osd_op_t*): Assertion `stripes[role].read_buf' failed. ```	2024-09-21 13:44:36 +03:00
Vitaliy Filippov	f18a749324	READ_CHAIN fix was incomplete :-)	2024-09-21 13:40:31 +03:00
Vitaliy Filippov	6e9307c522	Fix possible overflow in is_zero()	2024-09-21 13:40:10 +03:00
Vitaliy Filippov	99adbb9483	Release 1.9.1 Hotfixes for OpenNebula and upgrade hotfix for 1.7 - Fix deploy.vitastor, save.vitastor, restore.vitastor scripts not working for nodes other than master oned - Fix deploy.vitastor not working for VMs without Vitastor disks - Disable clearing old PG configuration when upgrading from 1.7 or older versions (it was breaking old clients)	2024-09-14 19:17:30 +03:00
Vitaliy Filippov	b489a611a9	Add 1.8 upgrade note	2024-09-14 19:17:30 +03:00
Vitaliy Filippov	c6c0b8957a	Stop updating old PG configuration when the user manually deletes it	2024-09-14 19:15:40 +03:00
Vitaliy Filippov	5d40d2a459	Fix oned.conf patch	2024-09-14 19:08:44 +03:00
Vitaliy Filippov	f449c28c3b	Always write decoded base64 deployment file (otherwise it breaks VMs without Vitastor disks)	2024-09-14 15:25:02 +03:00
Vitaliy Filippov	a6274f58cc	Same fix for save/restore: they also need to ssh to target node	2024-09-14 02:46:48 +03:00
Vitaliy Filippov	ac29ffea6a	Add ssh to target node to deploy.vitastor - without it it always tried to deploy VMs on oned host	2024-09-14 02:15:24 +03:00
Vitaliy Filippov	bc06acc153	Disable clearing old PG configuration - we can not be sure that old clients do not need it	2024-09-13 19:00:12 +03:00
Vitaliy Filippov	fe8e611e23	Release 1.9.0 - OpenNebula support! [Installation instructions](https://git.yourcmc.ru/vitalif/vitastor/src/branch/master/docs/installation/opennebula.en.md) - Added [vitastor-cli rm --exact\|--matching](https://git.yourcmc.ru/vitalif/vitastor/src/branch/master/docs/usage/cli.en.md#rm) command - Added [vitastor-cli dd](https://git.yourcmc.ru/vitalif/vitastor/src/branch/master/docs/usage/cli.en.md#dd) command - copy files between Vitastor images, files and pipes - Add a startup timeout to vitastor-cli to not wait for etcd infinitely - Fix non-working OSD_OP_READ_CHAIN_BITMAP O_o - Autodetect block_size/bitmap_granularity/immediate_commit when creating pools - Do not allow to create multiple pools with the same name from vitastor-cli - Fix skip_cache_check option not applied due to type issue (see github issue #70)	2024-09-06 01:46:16 +03:00
Vitaliy Filippov	7636f9c726	Turn off brp-python-bytecompile in RPM specs	2024-09-06 01:44:44 +03:00
ace	b85dab8583	use fio 3.35-1 for AlmaLinux 9	2024-05-18 21:17:16 +03:00