Release 1.4.5

- Fix a write stall caused by incorrect journal trimming introduced in 1.4.4 :)
- Fix PGs sometimes hanging in "starting" state on mass OSD restarts
- Fix a rare crash with "map::at" during OSD pings
- Use new defaults for non-capacitor (desktop) SSDs - improves T1Q256 random write from ~6k iops to ~45k iops
- Make journal_trim_interval configurable

.gitea/workflows/test.yml

@@ -395,7 +395,7 @@ jobs:
     steps:
     - name: Run test
       id: test
-      timeout-minutes: 3
+      timeout-minutes: 6
       run: SCHEME=ec /root/vitastor/tests/test_snapshot_chain.sh
     - name: Print logs
       if: always() && steps.test.outcome == 'failure'
@@ -532,6 +532,24 @@ jobs:
           echo ""
         done
 
+  test_switch_primary:
+    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_switch_primary.sh
+    - name: Print logs
+      if: always() && steps.test.outcome == 'failure'
+      run: |
+        for i in /root/vitastor/testdata/*.log /root/vitastor/testdata/*.txt; do
+          echo "-------- $i --------"
+          cat $i
+          echo ""
+        done
+
   test_write:
     runs-on: ubuntu-latest
     needs: build

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

@@ -39,6 +39,10 @@ for my $line (<>)
                 $test_name .= '_'.lc($1).'_'.$2;
             }
         }
+        if ($test_name eq 'test_snapshot_chain_ec')
+        {
+            $timeout = 6;
+        }
         $line =~ s!\./test_!/root/vitastor/tests/test_!;
         # Gitea CI doesn't support artifacts yet, lol
         #- name: Upload results

CLA-en.md

@@ -38,7 +38,7 @@ in the offer.
         on behalf of third parties, including on behalf of his employer.
 
 2. Subject of the Agreement. \
-   2.1. Subject of the Agreement shall be the Contributions sent to the Author by Contributors.
+   2.1. Subject of the Agreement shall be the Contributions sent to the Author by Contributors. \
    2.2. The Contributor grants to the Author the right to use Contributions at his own
         discretion and without any necessity to get a prior approval from Contributor or
         any other third party in any way, under a simple (non-exclusive), royalty-free,
@@ -86,7 +86,7 @@ in the offer.
         of their provision to the Author. \
    5.2. The Contributor represents and warrants that he legally owns exclusive
         intellectual property rights to the Contributions. \
-   5.3. The Contributor represents and warrants that any further use of \
+   5.3. The Contributor represents and warrants that any further use of
         Contributions by the Author as provided by Contributor under the terms
         of the Agreement does not infringe on intellectual and other rights and
         legitimate interests of third parties. \

CMakeLists.txt

@@ -2,6 +2,6 @@ cmake_minimum_required(VERSION 2.8.12)
 
 project(vitastor)
 
-set(VERSION "1.3.1")
+set(VERSION "1.4.5")
 
 add_subdirectory(src)

csi/Makefile

@@ -1,4 +1,4 @@
-VERSION ?= v1.3.1
+VERSION ?= v1.4.5
 
 all: build push
 

csi/deploy/004-csi-nodeplugin.yaml

@@ -49,7 +49,7 @@ spec:
             capabilities:
               add: ["SYS_ADMIN"]
             allowPrivilegeEscalation: true
-          image: vitalif/vitastor-csi:v1.3.1
+          image: vitalif/vitastor-csi:v1.4.5
           args:
             - "--node=$(NODE_ID)"
             - "--endpoint=$(CSI_ENDPOINT)"

csi/deploy/007-csi-provisioner.yaml

@@ -121,7 +121,7 @@ spec:
             privileged: true
             capabilities:
               add: ["SYS_ADMIN"]
-          image: vitalif/vitastor-csi:v1.3.1
+          image: vitalif/vitastor-csi:v1.4.5
           args:
             - "--node=$(NODE_ID)"
             - "--endpoint=$(CSI_ENDPOINT)"

csi/src/config.go

@@ -5,7 +5,7 @@ package vitastor
 
 const (
     vitastorCSIDriverName    = "csi.vitastor.io"
-    vitastorCSIDriverVersion = "1.3.1"
+    vitastorCSIDriverVersion = "1.4.5"
 )
 
 // Config struct fills the parameters of request or user input

csi/src/nodeserver.go

@@ -188,7 +188,6 @@ func (ns *NodeServer) unmapNbd(devicePath string)
 
 func findByPidFile(pidFile string) (*os.Process, error)
 {
-    klog.Infof("killing process with PID from file %s", pidFile)
     pidBuf, err := os.ReadFile(pidFile)
     if (err != nil)
     {
@@ -209,6 +208,7 @@ func findByPidFile(pidFile string) (*os.Process, error)
 
 func killByPidFile(pidFile string) error
 {
+    klog.Infof("killing process with PID from file %s", pidFile)
     proc, err := findByPidFile(pidFile)
     if (err != nil)
     {

debian/changelog

@@ -1,4 +1,4 @@
-vitastor (1.3.1-1) unstable; urgency=medium
+vitastor (1.4.5-1) unstable; urgency=medium
 
   * Bugfixes
 

debian/vitastor.Dockerfile

@@ -35,8 +35,8 @@ RUN set -e -x; \
     mkdir -p /root/packages/vitastor-$REL; \
     rm -rf /root/packages/vitastor-$REL/*; \
     cd /root/packages/vitastor-$REL; \
-    cp -r /root/vitastor vitastor-1.3.1; \
-    cd vitastor-1.3.1; \
+    cp -r /root/vitastor vitastor-1.4.5; \
+    cd vitastor-1.4.5; \
     ln -s /root/fio-build/fio-*/ ./fio; \
     FIO=$(head -n1 fio/debian/changelog | perl -pe 's/^.*\((.*?)\).*$/$1/'); \
     ls /usr/include/linux/raw.h || cp ./debian/raw.h /usr/include/linux/raw.h; \
@@ -49,8 +49,8 @@ RUN set -e -x; \
     rm -rf a b; \
     echo "dep:fio=$FIO" > debian/fio_version; \
     cd /root/packages/vitastor-$REL; \
-    tar --sort=name --mtime='2020-01-01' --owner=0 --group=0 --exclude=debian -cJf vitastor_1.3.1.orig.tar.xz vitastor-1.3.1; \
-    cd vitastor-1.3.1; \
+    tar --sort=name --mtime='2020-01-01' --owner=0 --group=0 --exclude=debian -cJf vitastor_1.4.5.orig.tar.xz vitastor-1.4.5; \
+    cd vitastor-1.4.5; \
     V=$(head -n1 debian/changelog | perl -pe 's/^.*\((.*?)\).*$/$1/'); \
     DEBFULLNAME="Vitaliy Filippov <vitalif@yourcmc.ru>" dch -D $REL -v "$V""$REL" "Rebuild for $REL"; \
     DEB_BUILD_OPTIONS=nocheck dpkg-buildpackage --jobs=auto -sa; \

docs/config/monitor.en.md

@@ -19,8 +19,8 @@ These parameters only apply to Monitors.
 ## etcd_mon_ttl
 
 - Type: seconds
-- Default: 30
-- Minimum: 10
+- Default: 1
+- Minimum: 5
 
 Monitor etcd lease refresh interval in seconds
 

docs/config/monitor.ru.md

@@ -19,8 +19,8 @@
 ## etcd_mon_ttl
 
 - Тип: секунды
-- Значение по умолчанию: 30
-- Минимальное значение: 10
+- Значение по умолчанию: 1
+- Минимальное значение: 5
 
 Интервал обновления etcd резервации (lease) монитором
 

docs/config/network.en.md

@@ -215,8 +215,8 @@ is scheduled.
 ## up_wait_retry_interval
 
 - Type: milliseconds
-- Default: 500
-- Minimum: 50
+- Default: 50
+- Minimum: 10
 - Can be changed online: yes
 
 OSDs respond to clients with a special error code when they receive I/O

docs/config/network.ru.md

@@ -224,8 +224,8 @@ OSD в любом случае согласовывают реальное зн
 ## up_wait_retry_interval
 
 - Тип: миллисекунды
-- Значение по умолчанию: 500
-- Минимальное значение: 50
+- Значение по умолчанию: 50
+- Минимальное значение: 10
 - Можно менять на лету: да
 
 Когда OSD получают от клиентов запросы ввода-вывода, относящиеся к не

docs/config/osd.en.md

@@ -59,6 +59,7 @@ them, even without restarting by updating configuration in etcd.
 - [recovery_tune_client_util_high](#recovery_tune_client_util_high)
 - [recovery_tune_agg_interval](#recovery_tune_agg_interval)
 - [recovery_tune_sleep_min_us](#recovery_tune_sleep_min_us)
+- [recovery_tune_sleep_cutoff_us](#recovery_tune_sleep_cutoff_us)
 
 ## etcd_report_interval
 
@@ -604,5 +605,14 @@ is usually fine.
 - Default: 10
 - Can be changed online: yes
 
-Minimum possible value for auto-tuned recovery_sleep_us. Values lower
-than this value are changed to 0.
+Minimum possible value for auto-tuned recovery_sleep_us. Lower values
+are changed to 0.
+
+## recovery_tune_sleep_cutoff_us
+
+- Type: microseconds
+- Default: 10000000
+- Can be changed online: yes
+
+Maximum possible value for auto-tuned recovery_sleep_us. Higher values
+are treated as outliers and ignored in aggregation.

docs/config/osd.ru.md

@@ -60,6 +60,7 @@
 - [recovery_tune_client_util_high](#recovery_tune_client_util_high)
 - [recovery_tune_agg_interval](#recovery_tune_agg_interval)
 - [recovery_tune_sleep_min_us](#recovery_tune_sleep_min_us)
+- [recovery_tune_sleep_cutoff_us](#recovery_tune_sleep_cutoff_us)
 
 ## etcd_report_interval
 
@@ -634,4 +635,14 @@ EC (кодов коррекции ошибок) с более, чем 1 диск
 - Можно менять на лету: да
 
 Минимальное возможное значение авто-подстроенного recovery_sleep_us.
-Значения ниже данного заменяются на 0.
+Меньшие значения заменяются на 0.
+
+## recovery_tune_sleep_cutoff_us
+
+- Тип: микросекунды
+- Значение по умолчанию: 10000000
+- Можно менять на лету: да
+
+Максимальное возможное значение авто-подстроенного recovery_sleep_us.
+Большие значения считаются случайными выбросами и игнорируются в
+усреднении.

docs/config/pool.en.md

@@ -154,6 +154,9 @@ That is, if it becomes impossible to place PG data on at least (pg_minsize)
 OSDs, PG is deactivated for both read and write. So you know that a fresh
 write always goes to at least (pg_minsize) OSDs (disks).
 
+That is, pg_size minus pg_minsize sets the number of disk failures to tolerate
+without temporary downtime (for [osd_out_time](monitor.en.md#osd_out_time)).
+
 FIXME: pg_minsize behaviour may be changed in the future to only make PGs
 read-only instead of deactivating them.
 

docs/config/pool.ru.md

@@ -157,6 +157,10 @@
 OSD, PG деактивируется на чтение и запись. Иными словами, всегда известно,
 что новые блоки данных всегда записываются как минимум на pg_minsize дисков.
 
+По сути, разница pg_size и pg_minsize задаёт число отказов дисков, которые пул
+может пережить без временной (на [osd_out_time](monitor.ru.md#osd_out_time))
+остановки обслуживания.
+
 FIXME: Поведение pg_minsize может быть изменено в будущем с полной деактивации
 PG на перевод их в режим только для чтения.
 

docs/config/src/monitor.yml

@@ -1,7 +1,7 @@
 - name: etcd_mon_ttl
   type: sec
-  min: 10
-  default: 30
+  min: 5
+  default: 1
   info: Monitor etcd lease refresh interval in seconds
   info_ru: Интервал обновления etcd резервации (lease) монитором
 - name: etcd_mon_timeout

docs/config/src/network.yml

@@ -245,8 +245,8 @@
     повторная попытка соединения.
 - name: up_wait_retry_interval
   type: ms
-  min: 50
-  default: 500
+  min: 10
+  default: 50
   online: true
   info: |
     OSDs respond to clients with a special error code when they receive I/O

docs/config/src/osd.yml

@@ -731,8 +731,19 @@
   default: 10
   online: true
   info: |
-    Minimum possible value for auto-tuned recovery_sleep_us. Values lower
-    than this value are changed to 0.
+    Minimum possible value for auto-tuned recovery_sleep_us. Lower values
+    are changed to 0.
   info_ru: |
     Минимальное возможное значение авто-подстроенного recovery_sleep_us.
-    Значения ниже данного заменяются на 0.
+    Меньшие значения заменяются на 0.
+- name: recovery_tune_sleep_cutoff_us
+  type: us
+  default: 10000000
+  online: true
+  info: |
+    Maximum possible value for auto-tuned recovery_sleep_us. Higher values
+    are treated as outliers and ignored in aggregation.
+  info_ru: |
+    Максимальное возможное значение авто-подстроенного recovery_sleep_us.
+    Большие значения считаются случайными выбросами и игнорируются в
+    усреднении.

docs/installation/kubernetes.en.md

@@ -37,6 +37,7 @@ Vitastor CSI supports:
 - 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
 - Upgrades with VDUSE - new handler processes are restarted when CSI pods are restarted themselves
+- VDUSE daemon auto-restart - handler processes are automatically restarted if they crash due to a bug in Vitastor client code
 - Multiple clusters by using multiple configuration files in ConfigMap.
 
 Remember that to use snapshots with CSI you also have to install [Snapshot Controller and CRDs](https://kubernetes-csi.github.io/docs/snapshot-controller.html#deployment).

docs/installation/kubernetes.ru.md

@@ -37,6 +37,7 @@ CSI-плагин Vitastor поддерживает:
 - 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)
 - Обновление при использовании VDUSE - новые процессы-обработчики устройств успешно перезапускаются вместе с самими подами CSI
+- Автоперезауск демонов VDUSE - процесс-обработчик автоматически перезапустится, если он внезапно упадёт из-за бага в коде клиента Vitastor
 - Несколько кластеров через задание нескольких файлов конфигурации в ConfigMap.
 
 Не забывайте, что для использования снимков нужно сначала установить [контроллер снимков и CRD](https://kubernetes-csi.github.io/docs/snapshot-controller.html#deployment).

docs/performance/theoretical.en.md

@@ -11,19 +11,26 @@ Replicated setups:
 - Single-threaded write+fsync latency:
   - With immediate commit: 2 network roundtrips + 1 disk write.
   - With lazy commit: 4 network roundtrips + 1 disk write + 1 disk flush.
-- Saturated parallel read iops: min(network bandwidth, sum(disk read iops)).
-- Saturated parallel write iops: min(network bandwidth, sum(disk write iops / number of replicas / write amplification)).
+- Linear read: `min(total network bandwidth, sum(disk read MB/s))`.
+- Linear write: `min(total network bandwidth, sum(disk write MB/s / number of replicas))`.
+- Saturated parallel read iops: `min(total network bandwidth, sum(disk read iops))`.
+- Saturated parallel write iops: `min(total network bandwidth / number of replicas, sum(disk write iops / number of replicas / (write amplification = 4)))`.
 
-EC/XOR setups:
+EC/XOR setups (EC N+K):
 - Single-threaded (T1Q1) read latency: 1.5 network roundtrips + 1 disk read.
 - Single-threaded write+fsync latency:
   - With immediate commit: 3.5 network roundtrips + 1 disk read + 2 disk writes.
   - With lazy commit: 5.5 network roundtrips + 1 disk read + 2 disk writes + 2 disk fsyncs.
-  - 0.5 in actually (k-1)/k which means that an additional roundtrip doesn't happen when
+  - 0.5 in actually `(N-1)/N` which means that an additional roundtrip doesn't happen when
     the read sub-operation can be served locally.
-- Saturated parallel read iops: min(network bandwidth, sum(disk read iops)).
-- Saturated parallel write iops: min(network bandwidth, sum(disk write iops * number of data drives / (number of data + parity drives) / write amplification)).
-  In fact, you should put disk write iops under the condition of ~10% reads / ~90% writes in this formula.
+- Linear read: `min(total network bandwidth, sum(disk read MB/s))`.
+- Linear write: `min(total network bandwidth, sum(disk write MB/s * N/(N+K)))`.
+- Saturated parallel read iops: `min(total network bandwidth, sum(disk read iops))`.
+- Saturated parallel write iops: roughly `total iops / (N+K) / WA`. More exactly,
+  `min(total network bandwidth * N/(N+K), sum(disk randrw iops / (N*4 + K*5 + 1)))` with
+  random read/write mix corresponding to `(N-1)/(N*4 + K*5 + 1)*100 % reads`.
+  - For example, with EC 2+1 it is: `(7% randrw iops) / 14`.
+  - With EC 6+3 it is: `(12.5% randrw iops) / 40`.
 
 Write amplification for 4 KB blocks is usually 3-5 in Vitastor:
 1. Journal block write

docs/performance/theoretical.ru.md

@@ -11,20 +11,27 @@
 - Запись+fsync в 1 поток:
   - С мгновенным сбросом: 2 RTT + 1 запись.
   - С отложенным ("ленивым") сбросом: 4 RTT + 1 запись + 1 fsync.
-- Параллельное чтение: сумма IOPS всех дисков либо производительность сети, если в сеть упрётся раньше.
-- Параллельная запись: сумма IOPS всех дисков / число реплик / WA либо производительность сети, если в сеть упрётся раньше.
+- Линейное чтение: сумма МБ/с чтения всех дисков, либо общая производительность сети (сумма пропускной способности сети всех нод), если в сеть упрётся раньше.
+- Линейная запись: сумма МБ/с записи всех дисков / число реплик, либо производительность сети / число реплик, если в сеть упрётся раньше.
+- Параллельное случайное мелкое чтение: сумма IOPS чтения всех дисков, либо производительность сети, если в сеть упрётся раньше.
+- Параллельная случайная мелкая запись: сумма IOPS записи всех дисков / число реплик / WA, либо производительность сети / число реплик, если в сеть упрётся раньше.
 
-При использовании кодов коррекции ошибок (EC):
+При использовании кодов коррекции ошибок (EC N+K):
 - Задержка чтения в 1 поток (T1Q1): 1.5 RTT + 1 чтение.
 - Запись+fsync в 1 поток:
   - С мгновенным сбросом: 3.5 RTT + 1 чтение + 2 записи.
   - С отложенным ("ленивым") сбросом: 5.5 RTT + 1 чтение + 2 записи + 2 fsync.
-- Под 0.5 на самом деле подразумевается (k-1)/k, где k - число дисков данных,
+- Под 0.5 на самом деле подразумевается (N-1)/N, где N - число дисков данных,
   что означает, что дополнительное обращение по сети не нужно, когда операция
   чтения обслуживается локально.
-- Параллельное чтение: сумма IOPS всех дисков либо производительность сети, если в сеть упрётся раньше.
-- Параллельная запись: сумма IOPS всех дисков / общее число дисков данных и чётности / WA либо производительность сети, если в сеть упрётся раньше.
-  Примечание: IOPS дисков в данном случае надо брать в смешанном режиме чтения/записи в пропорции, аналогичной формулам выше.
+- Линейное чтение: сумма МБ/с чтения всех дисков, либо общая производительность сети, если в сеть упрётся раньше.
+- Линейная запись: сумма МБ/с записи всех дисков * N/(N+K), либо производительность сети * N / (N+K), если в сеть упрётся раньше.
+- Параллельное случайное мелкое чтение: сумма IOPS чтения всех дисков либо производительность сети, если в сеть упрётся раньше.
+- Параллельная случайная мелкая запись: грубо `(сумма IOPS / (N+K) / WA)`. Если точнее, то:
+  сумма смешанного IOPS всех дисков при `(N-1)/(N*4 + K*5 + 1)*100 %` чтения, делённая на `(N*4 + K*5 + 1)`.
+  Либо, производительность сети * N/(N+K), если в сеть упрётся раньше.
+  - Например, при EC 2+1 это: `(сумма IOPS при 7% чтения) / 14`.
+  - При EC 6+3 это: `(сумма IOPS при 12.5% чтения) / 40`.
 
 WA (мультипликатор записи) для 4 КБ блоков в Vitastor обычно составляет 3-5:
 1. Запись метаданных в журнал

mon/mon.js

@@ -55,7 +55,7 @@ const etcd_tree = {
             // etcd connection - configurable online
             etcd_address: "10.0.115.10:2379/v3",
             // mon
-            etcd_mon_ttl: 30, // min: 10
+            etcd_mon_ttl: 5, // min: 1
             etcd_mon_timeout: 1000, // ms. min: 0
             etcd_mon_retries: 5, // min: 0
             mon_change_timeout: 1000, // ms. min: 100
@@ -92,7 +92,7 @@ const etcd_tree = {
             peer_connect_timeout: 5, // seconds. min: 1
             osd_idle_timeout: 5, // seconds. min: 1
             osd_ping_timeout: 5, // seconds. min: 1
-            up_wait_retry_interval: 500, // ms. min: 50
+            up_wait_retry_interval: 50, // ms. min: 10
             max_etcd_attempts: 5,
             etcd_quick_timeout: 1000, // ms
             etcd_slow_timeout: 5000, // ms
@@ -390,7 +390,8 @@ class Mon
 {
     constructor(config)
     {
-        this.die = (e) => this._die(e);
+        this.failconnect = (e) => this._die(e, 2);
+        this.die = (e) => this._die(e, 1);
         if (fs.existsSync(config.config_path||'/etc/vitastor/vitastor.conf'))
         {
             config = {
@@ -479,10 +480,10 @@ class Mon
 
     check_config()
     {
-        this.config.etcd_mon_ttl = Number(this.config.etcd_mon_ttl) || 30;
-        if (this.config.etcd_mon_ttl < 10)
+        this.config.etcd_mon_ttl = Number(this.config.etcd_mon_ttl) || 5;
+        if (this.config.etcd_mon_ttl < 1)
         {
-            this.config.etcd_mon_ttl = 10;
+            this.config.etcd_mon_ttl = 1;
         }
         this.config.etcd_mon_timeout = Number(this.config.etcd_mon_timeout) || 0;
         if (this.config.etcd_mon_timeout <= 0)
@@ -604,7 +605,7 @@ class Mon
         }
         if (!this.ws)
         {
-            this.die('Failed to open etcd watch websocket');
+            this.failconnect('Failed to open etcd watch websocket');
         }
         const cur_addr = this.selected_etcd_url;
         this.ws_alive = true;
@@ -674,7 +675,12 @@ class Mon
                 {
                     this.parse_kv(e.kv);
                     const key = e.kv.key.substr(this.etcd_prefix.length);
-                    if (key.substr(0, 11) == '/osd/stats/' || key.substr(0, 10) == '/pg/stats/' || key.substr(0, 16) == '/osd/inodestats/')
+                    if (key.substr(0, 11) == '/osd/state/')
+                    {
+                        stats_changed = true;
+                        changed = true;
+                    }
+                    else if (key.substr(0, 11) == '/osd/stats/' || key.substr(0, 10) == '/pg/stats/' || key.substr(0, 16) == '/osd/inodestats/')
                     {
                         stats_changed = true;
                     }
@@ -791,9 +797,9 @@ class Mon
             const res = await this.etcd_call('/lease/keepalive', { ID: this.etcd_lease_id }, this.config.etcd_mon_timeout, this.config.etcd_mon_retries);
             if (!res.result.TTL)
             {
-                this.die('Lease expired');
+                this.failconnect('Lease expired');
             }
-        }, this.config.etcd_mon_timeout);
+        }, this.config.etcd_mon_ttl*1000);
         if (!this.signals_set)
         {
             process.on('SIGINT', this.on_stop_cb);
@@ -1414,7 +1420,14 @@ class Mon
             }
             if (changed)
             {
-                await this.save_pg_config(new_config_pgs);
+                const ok = await this.save_pg_config(new_config_pgs);
+                if (ok)
+                    console.log('PG configuration successfully changed');
+                else
+                {
+                    console.log('Someone changed PG configuration while we also tried to change it. Retrying in '+this.config.mon_change_timeout+' ms');
+                    this.schedule_recheck();
+                }
             }
         }
         this.recheck_pgs_active = false;
@@ -1495,6 +1508,11 @@ class Mon
             this.save_new_pgs_txn(new_config_pgs, etcd_request, pool_id, up_osds, osd_tree, real_prev_pgs, pool_res.pgs, pg_history);
         }
         new_config_pgs.hash = tree_hash;
+        return await this.save_pg_config(new_config_pgs, etcd_request);
+    }
+
+    async save_pg_config(new_config_pgs, etcd_request = { compare: [], success: [] })
+    {
         etcd_request.compare.push(
             { key: b64(this.etcd_prefix+'/mon/master'), target: 'LEASE', lease: ''+this.etcd_lease_id },
             { key: b64(this.etcd_prefix+'/config/pgs'), target: 'MOD', mod_revision: ''+this.etcd_watch_revision, result: 'LESS' },
@@ -1622,9 +1640,13 @@ class Mon
         }
         const sum_diff = { op_stats: {}, subop_stats: {}, recovery_stats: {} };
         // Sum derived values instead of deriving summed
-        for (const osd in this.state.osd.stats)
+        for (const osd in this.state.osd.state)
         {
             const derived = this.prev_stats.osd_diff[osd];
+            if (!this.state.osd.state[osd] || !derived)
+            {
+                continue;
+            }
             for (const type in sum_diff)
             {
                 for (const op in derived[type]||{})
@@ -1725,9 +1747,13 @@ class Mon
             const used = this.state.pool.stats[pool_id].used_raw_tb;
             this.state.pool.stats[pool_id].used_raw_tb = Number(used)/1024/1024/1024/1024;
         }
-        for (const osd_num in this.state.osd.inodestats)
+        for (const osd_num in this.state.osd.state)
         {
             const ist = this.state.osd.inodestats[osd_num];
+            if (!ist || !this.state.osd.state[osd_num])
+            {
+                continue;
+            }
             for (const pool_id in ist)
             {
                 inode_stats[pool_id] = inode_stats[pool_id] || {};
@@ -1743,9 +1769,14 @@ class Mon
                 }
             }
         }
-        for (const osd in this.prev_stats.osd_diff)
+        for (const osd in this.state.osd.state)
         {
-            for (const pool_id in this.prev_stats.osd_diff[osd].inode_stats)
+            const osd_diff = this.prev_stats.osd_diff[osd];
+            if (!osd_diff || !this.state.osd.state[osd])
+            {
+                continue;
+            }
+            for (const pool_id in osd_diff.inode_stats)
             {
                 for (const inode_num in this.prev_stats.osd_diff[osd].inode_stats[pool_id])
                 {
@@ -1985,14 +2016,14 @@ class Mon
                 return res.json;
             }
         }
-        this.die();
+        this.failconnect();
     }
 
-    _die(err)
+    _die(err, code)
     {
         // In fact we can just try to rejoin
         console.error(new Error(err || 'Cluster connection failed'));
-        process.exit(1);
+        process.exit(code || 2);
     }
 
     local_ips(all)

mon/package.json

@@ -1,6 +1,6 @@
 {
   "name": "vitastor-mon",
-  "version": "1.3.1",
+  "version": "1.4.5",
   "description": "Vitastor SDS monitor service",
   "main": "mon-main.js",
   "scripts": {

mon/vitastor-osd@.service

@@ -8,7 +8,9 @@ PartOf=vitastor.target
 LimitNOFILE=1048576
 LimitNPROC=1048576
 LimitMEMLOCK=infinity
-ExecStart=bash -c 'exec vitastor-disk exec-osd /dev/vitastor/osd%i-data >>/var/log/vitastor/osd%i.log 2>&1'
+# Use the following for direct logs to files
+#ExecStart=bash -c 'exec vitastor-disk exec-osd /dev/vitastor/osd%i-data >>/var/log/vitastor/osd%i.log 2>&1'
+ExecStart=vitastor-disk exec-osd /dev/vitastor/osd%i-data
 ExecStartPre=+vitastor-disk pre-exec /dev/vitastor/osd%i-data
 WorkingDirectory=/
 User=vitastor

patches/cinder-vitastor.py

@@ -50,7 +50,7 @@ from cinder.volume import configuration
 from cinder.volume import driver
 from cinder.volume import volume_utils
 
-VERSION = '1.3.1'
+VERSION = '1.4.5'
 
 LOG = logging.getLogger(__name__)
 

patches/libvirt-9.10-vitastor.diff

@@ -0,0 +1,643 @@
+commit c1cd026e211e94b120028e7c98a6e4ce5afe9846
+Author: Vitaliy Filippov <vitalif@yourcmc.ru>
+Date:   Wed Jan 24 22:04:50 2024 +0300
+
+    Add Vitastor support
+
+diff --git a/include/libvirt/libvirt-storage.h b/include/libvirt/libvirt-storage.h
+index aaad4a3da1..5f5daa8341 100644
+--- a/include/libvirt/libvirt-storage.h
++++ b/include/libvirt/libvirt-storage.h
+@@ -326,6 +326,7 @@ typedef enum {
+     VIR_CONNECT_LIST_STORAGE_POOLS_ZFS           = 1 << 17, /* (Since: 1.2.8) */
+     VIR_CONNECT_LIST_STORAGE_POOLS_VSTORAGE      = 1 << 18, /* (Since: 3.1.0) */
+     VIR_CONNECT_LIST_STORAGE_POOLS_ISCSI_DIRECT  = 1 << 19, /* (Since: 5.6.0) */
++    VIR_CONNECT_LIST_STORAGE_POOLS_VITASTOR      = 1 << 20, /* (Since: 5.0.0) */
+ } virConnectListAllStoragePoolsFlags;
+ 
+ int                     virConnectListAllStoragePools(virConnectPtr conn,
+diff --git a/src/conf/domain_conf.c b/src/conf/domain_conf.c
+index 22ad43e1d7..56c81d6852 100644
+--- a/src/conf/domain_conf.c
++++ b/src/conf/domain_conf.c
+@@ -7185,7 +7185,8 @@ virDomainDiskSourceNetworkParse(xmlNodePtr node,
+     src->configFile = virXPathString("string(./config/@file)", ctxt);
+ 
+     if (src->protocol == VIR_STORAGE_NET_PROTOCOL_HTTP ||
+-        src->protocol == VIR_STORAGE_NET_PROTOCOL_HTTPS)
++        src->protocol == VIR_STORAGE_NET_PROTOCOL_HTTPS ||
++        src->protocol == VIR_STORAGE_NET_PROTOCOL_VITASTOR)
+         src->query = virXMLPropString(node, "query");
+ 
+     if (virDomainStorageNetworkParseHosts(node, ctxt, &src->hosts, &src->nhosts) < 0)
+@@ -30618,6 +30619,7 @@ virDomainStorageSourceTranslateSourcePool(virStorageSource *src,
+ 
+     case VIR_STORAGE_POOL_MPATH:
+     case VIR_STORAGE_POOL_RBD:
++    case VIR_STORAGE_POOL_VITASTOR:
+     case VIR_STORAGE_POOL_SHEEPDOG:
+     case VIR_STORAGE_POOL_GLUSTER:
+     case VIR_STORAGE_POOL_LAST:
+diff --git a/src/conf/domain_validate.c b/src/conf/domain_validate.c
+index c72108886e..c739ed6c43 100644
+--- a/src/conf/domain_validate.c
++++ b/src/conf/domain_validate.c
+@@ -495,6 +495,7 @@ virDomainDiskDefValidateSourceChainOne(const virStorageSource *src)
+         case VIR_STORAGE_NET_PROTOCOL_RBD:
+             break;
+ 
++        case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
+         case VIR_STORAGE_NET_PROTOCOL_NBD:
+         case VIR_STORAGE_NET_PROTOCOL_SHEEPDOG:
+         case VIR_STORAGE_NET_PROTOCOL_GLUSTER:
+@@ -541,7 +542,7 @@ virDomainDiskDefValidateSourceChainOne(const virStorageSource *src)
+         }
+     }
+ 
+-    /* internal snapshots and config files are currently supported only with rbd: */
++    /* internal snapshots are currently supported only with rbd: */
+     if (virStorageSourceGetActualType(src) != VIR_STORAGE_TYPE_NETWORK &&
+         src->protocol != VIR_STORAGE_NET_PROTOCOL_RBD) {
+         if (src->snapshot) {
+@@ -549,10 +550,15 @@ virDomainDiskDefValidateSourceChainOne(const virStorageSource *src)
+                            _("<snapshot> element is currently supported only with 'rbd' disks"));
+             return -1;
+         }
++    }
+ 
++    /* config files are currently supported only with rbd and vitastor: */
++    if (virStorageSourceGetActualType(src) != VIR_STORAGE_TYPE_NETWORK &&
++        src->protocol != VIR_STORAGE_NET_PROTOCOL_RBD &&
++        src->protocol != VIR_STORAGE_NET_PROTOCOL_VITASTOR) {
+         if (src->configFile) {
+             virReportError(VIR_ERR_XML_ERROR, "%s",
+-                           _("<config> element is currently supported only with 'rbd' disks"));
++                           _("<config> element is currently supported only with 'rbd' and 'vitastor' disks"));
+             return -1;
+         }
+     }
+diff --git a/src/conf/schemas/domaincommon.rng b/src/conf/schemas/domaincommon.rng
+index b98a2ae602..7d7a872e01 100644
+--- a/src/conf/schemas/domaincommon.rng
++++ b/src/conf/schemas/domaincommon.rng
+@@ -1997,6 +1997,35 @@
+     </element>
+   </define>
+ 
++  <define name="diskSourceNetworkProtocolVitastor">
++    <element name="source">
++      <interleave>
++        <attribute name="protocol">
++          <value>vitastor</value>
++        </attribute>
++        <ref name="diskSourceCommon"/>
++        <optional>
++          <attribute name="name"/>
++        </optional>
++        <optional>
++          <attribute name="query"/>
++        </optional>
++        <zeroOrMore>
++          <ref name="diskSourceNetworkHost"/>
++        </zeroOrMore>
++        <optional>
++          <element name="config">
++            <attribute name="file">
++              <ref name="absFilePath"/>
++            </attribute>
++            <empty/>
++          </element>
++        </optional>
++        <empty/>
++      </interleave>
++    </element>
++  </define>
++
+   <define name="diskSourceNetworkProtocolISCSI">
+     <element name="source">
+       <attribute name="protocol">
+@@ -2347,6 +2376,7 @@
+       <ref name="diskSourceNetworkProtocolSimple"/>
+       <ref name="diskSourceNetworkProtocolVxHS"/>
+       <ref name="diskSourceNetworkProtocolNFS"/>
++      <ref name="diskSourceNetworkProtocolVitastor"/>
+     </choice>
+   </define>
+ 
+diff --git a/src/conf/storage_conf.c b/src/conf/storage_conf.c
+index 68842004b7..1d69a788b6 100644
+--- a/src/conf/storage_conf.c
++++ b/src/conf/storage_conf.c
+@@ -56,7 +56,7 @@ VIR_ENUM_IMPL(virStoragePool,
+               "logical", "disk", "iscsi",
+               "iscsi-direct", "scsi", "mpath",
+               "rbd", "sheepdog", "gluster",
+-              "zfs", "vstorage",
++              "zfs", "vstorage", "vitastor",
+ );
+ 
+ VIR_ENUM_IMPL(virStoragePoolFormatFileSystem,
+@@ -242,6 +242,18 @@ static virStoragePoolTypeInfo poolTypeInfo[] = {
+           .formatToString = virStorageFileFormatTypeToString,
+       }
+     },
++    {.poolType = VIR_STORAGE_POOL_VITASTOR,
++     .poolOptions = {
++         .flags = (VIR_STORAGE_POOL_SOURCE_HOST |
++                   VIR_STORAGE_POOL_SOURCE_NETWORK |
++                   VIR_STORAGE_POOL_SOURCE_NAME),
++      },
++      .volOptions = {
++          .defaultFormat = VIR_STORAGE_FILE_RAW,
++          .formatFromString = virStorageVolumeFormatFromString,
++          .formatToString = virStorageFileFormatTypeToString,
++      }
++    },
+     {.poolType = VIR_STORAGE_POOL_SHEEPDOG,
+      .poolOptions = {
+          .flags = (VIR_STORAGE_POOL_SOURCE_HOST |
+@@ -538,6 +550,11 @@ virStoragePoolDefParseSource(xmlXPathContextPtr ctxt,
+                        _("element 'name' is mandatory for RBD pool"));
+         return -1;
+     }
++    if (pool_type == VIR_STORAGE_POOL_VITASTOR && source->name == NULL) {
++        virReportError(VIR_ERR_XML_ERROR, "%s",
++                       _("element 'name' is mandatory for Vitastor pool"));
++        return -1;
++    }
+ 
+     if (options->formatFromString) {
+         g_autofree char *format = NULL;
+@@ -1127,6 +1144,7 @@ virStoragePoolDefFormatBuf(virBuffer *buf,
+     /* RBD, Sheepdog, Gluster and Iscsi-direct devices are not local block devs nor
+      * files, so they don't have a target */
+     if (def->type != VIR_STORAGE_POOL_RBD &&
++        def->type != VIR_STORAGE_POOL_VITASTOR &&
+         def->type != VIR_STORAGE_POOL_SHEEPDOG &&
+         def->type != VIR_STORAGE_POOL_GLUSTER &&
+         def->type != VIR_STORAGE_POOL_ISCSI_DIRECT) {
+diff --git a/src/conf/storage_conf.h b/src/conf/storage_conf.h
+index fc67957cfe..720c07ef74 100644
+--- a/src/conf/storage_conf.h
++++ b/src/conf/storage_conf.h
+@@ -103,6 +103,7 @@ typedef enum {
+     VIR_STORAGE_POOL_GLUSTER,  /* Gluster device */
+     VIR_STORAGE_POOL_ZFS,      /* ZFS */
+     VIR_STORAGE_POOL_VSTORAGE, /* Virtuozzo Storage */
++    VIR_STORAGE_POOL_VITASTOR, /* Vitastor */
+ 
+     VIR_STORAGE_POOL_LAST,
+ } virStoragePoolType;
+@@ -454,6 +455,7 @@ VIR_ENUM_DECL(virStoragePartedFs);
+                  VIR_CONNECT_LIST_STORAGE_POOLS_SCSI     | \
+                  VIR_CONNECT_LIST_STORAGE_POOLS_MPATH    | \
+                  VIR_CONNECT_LIST_STORAGE_POOLS_RBD      | \
++                 VIR_CONNECT_LIST_STORAGE_POOLS_VITASTOR | \
+                  VIR_CONNECT_LIST_STORAGE_POOLS_SHEEPDOG | \
+                  VIR_CONNECT_LIST_STORAGE_POOLS_GLUSTER  | \
+                  VIR_CONNECT_LIST_STORAGE_POOLS_ZFS      | \
+diff --git a/src/conf/storage_source_conf.c b/src/conf/storage_source_conf.c
+index f974a521b1..cd394d0a9f 100644
+--- a/src/conf/storage_source_conf.c
++++ b/src/conf/storage_source_conf.c
+@@ -88,6 +88,7 @@ VIR_ENUM_IMPL(virStorageNetProtocol,
+               "ssh",
+               "vxhs",
+               "nfs",
++              "vitastor",
+ );
+ 
+ 
+@@ -1301,6 +1302,7 @@ virStorageSourceNetworkDefaultPort(virStorageNetProtocol protocol)
+         case VIR_STORAGE_NET_PROTOCOL_GLUSTER:
+             return 24007;
+ 
++        case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
+         case VIR_STORAGE_NET_PROTOCOL_RBD:
+             /* we don't provide a default for RBD */
+             return 0;
+diff --git a/src/conf/storage_source_conf.h b/src/conf/storage_source_conf.h
+index 5e7d127453..283709eeb3 100644
+--- a/src/conf/storage_source_conf.h
++++ b/src/conf/storage_source_conf.h
+@@ -129,6 +129,7 @@ typedef enum {
+     VIR_STORAGE_NET_PROTOCOL_SSH,
+     VIR_STORAGE_NET_PROTOCOL_VXHS,
+     VIR_STORAGE_NET_PROTOCOL_NFS,
++    VIR_STORAGE_NET_PROTOCOL_VITASTOR,
+ 
+     VIR_STORAGE_NET_PROTOCOL_LAST
+ } virStorageNetProtocol;
+diff --git a/src/conf/virstorageobj.c b/src/conf/virstorageobj.c
+index 59fa5da372..4739167f5f 100644
+--- a/src/conf/virstorageobj.c
++++ b/src/conf/virstorageobj.c
+@@ -1438,6 +1438,7 @@ virStoragePoolObjSourceFindDuplicateCb(const void *payload,
+             return 1;
+         break;
+ 
++    case VIR_STORAGE_POOL_VITASTOR:
+     case VIR_STORAGE_POOL_ISCSI_DIRECT:
+     case VIR_STORAGE_POOL_RBD:
+     case VIR_STORAGE_POOL_LAST:
+@@ -1921,6 +1922,8 @@ virStoragePoolObjMatch(virStoragePoolObj *obj,
+                (obj->def->type == VIR_STORAGE_POOL_MPATH))   ||
+               (MATCH(VIR_CONNECT_LIST_STORAGE_POOLS_RBD) &&
+                (obj->def->type == VIR_STORAGE_POOL_RBD))     ||
++              (MATCH(VIR_CONNECT_LIST_STORAGE_POOLS_VITASTOR) &&
++               (obj->def->type == VIR_STORAGE_POOL_VITASTOR)) ||
+               (MATCH(VIR_CONNECT_LIST_STORAGE_POOLS_SHEEPDOG) &&
+                (obj->def->type == VIR_STORAGE_POOL_SHEEPDOG)) ||
+               (MATCH(VIR_CONNECT_LIST_STORAGE_POOLS_GLUSTER) &&
+diff --git a/src/libvirt-storage.c b/src/libvirt-storage.c
+index db7660aac4..561df34709 100644
+--- a/src/libvirt-storage.c
++++ b/src/libvirt-storage.c
+@@ -94,6 +94,7 @@ virStoragePoolGetConnect(virStoragePoolPtr pool)
+  * VIR_CONNECT_LIST_STORAGE_POOLS_SCSI
+  * VIR_CONNECT_LIST_STORAGE_POOLS_MPATH
+  * VIR_CONNECT_LIST_STORAGE_POOLS_RBD
++ * VIR_CONNECT_LIST_STORAGE_POOLS_VITASTOR
+  * VIR_CONNECT_LIST_STORAGE_POOLS_SHEEPDOG
+  * VIR_CONNECT_LIST_STORAGE_POOLS_GLUSTER
+  * VIR_CONNECT_LIST_STORAGE_POOLS_ZFS
+diff --git a/src/libxl/libxl_conf.c b/src/libxl/libxl_conf.c
+index 62e1be6672..71a1d42896 100644
+--- a/src/libxl/libxl_conf.c
++++ b/src/libxl/libxl_conf.c
+@@ -979,6 +979,7 @@ libxlMakeNetworkDiskSrcStr(virStorageSource *src,
+     case VIR_STORAGE_NET_PROTOCOL_SSH:
+     case VIR_STORAGE_NET_PROTOCOL_VXHS:
+     case VIR_STORAGE_NET_PROTOCOL_NFS:
++    case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
+     case VIR_STORAGE_NET_PROTOCOL_LAST:
+     case VIR_STORAGE_NET_PROTOCOL_NONE:
+         virReportError(VIR_ERR_NO_SUPPORT,
+diff --git a/src/libxl/xen_xl.c b/src/libxl/xen_xl.c
+index f175359307..8efcf4c329 100644
+--- a/src/libxl/xen_xl.c
++++ b/src/libxl/xen_xl.c
+@@ -1456,6 +1456,7 @@ xenFormatXLDiskSrcNet(virStorageSource *src)
+     case VIR_STORAGE_NET_PROTOCOL_SSH:
+     case VIR_STORAGE_NET_PROTOCOL_VXHS:
+     case VIR_STORAGE_NET_PROTOCOL_NFS:
++    case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
+     case VIR_STORAGE_NET_PROTOCOL_LAST:
+     case VIR_STORAGE_NET_PROTOCOL_NONE:
+         virReportError(VIR_ERR_NO_SUPPORT,
+diff --git a/src/qemu/qemu_block.c b/src/qemu/qemu_block.c
+index 7e9daf0bdc..825b4a3006 100644
+--- a/src/qemu/qemu_block.c
++++ b/src/qemu/qemu_block.c
+@@ -758,6 +758,38 @@ qemuBlockStorageSourceGetRBDProps(virStorageSource *src,
+ }
+ 
+ 
++static virJSONValue *
++qemuBlockStorageSourceGetVitastorProps(virStorageSource *src)
++{
++    virJSONValue *ret = NULL;
++    virStorageNetHostDef *host;
++    size_t i;
++    g_auto(virBuffer) buf = VIR_BUFFER_INITIALIZER;
++    g_autofree char *etcd = NULL;
++
++    for (i = 0; i < src->nhosts; i++) {
++        host = src->hosts + i;
++        if ((virStorageNetHostTransport)host->transport != VIR_STORAGE_NET_HOST_TRANS_TCP) {
++            return NULL;
++        }
++        virBufferAsprintf(&buf, i > 0 ? ",%s:%u" : "%s:%u", host->name, host->port);
++    }
++    if (src->nhosts > 0) {
++        etcd = virBufferContentAndReset(&buf);
++    }
++
++    if (virJSONValueObjectAdd(&ret,
++                              "S:etcd-host", etcd,
++                              "S:etcd-prefix", src->query,
++                              "S:config-path", src->configFile,
++                              "s:image", src->path,
++                              NULL) < 0)
++        return NULL;
++
++    return ret;
++}
++
++
+ static virJSONValue *
+ qemuBlockStorageSourceGetSheepdogProps(virStorageSource *src)
+ {
+@@ -1140,6 +1172,12 @@ qemuBlockStorageSourceGetBackendProps(virStorageSource *src,
+                 return NULL;
+             break;
+ 
++        case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
++            driver = "vitastor";
++            if (!(fileprops = qemuBlockStorageSourceGetVitastorProps(src)))
++                return NULL;
++            break;
++
+         case VIR_STORAGE_NET_PROTOCOL_SHEEPDOG:
+             driver = "sheepdog";
+             if (!(fileprops = qemuBlockStorageSourceGetSheepdogProps(src)))
+@@ -2032,6 +2070,7 @@ qemuBlockGetBackingStoreString(virStorageSource *src,
+ 
+             case VIR_STORAGE_NET_PROTOCOL_SHEEPDOG:
+             case VIR_STORAGE_NET_PROTOCOL_RBD:
++            case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
+             case VIR_STORAGE_NET_PROTOCOL_VXHS:
+             case VIR_STORAGE_NET_PROTOCOL_NFS:
+             case VIR_STORAGE_NET_PROTOCOL_SSH:
+@@ -2415,6 +2454,12 @@ qemuBlockStorageSourceCreateGetStorageProps(virStorageSource *src,
+                 return -1;
+             break;
+ 
++        case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
++            driver = "vitastor";
++            if (!(location = qemuBlockStorageSourceGetVitastorProps(src)))
++                return -1;
++            break;
++
+         case VIR_STORAGE_NET_PROTOCOL_SHEEPDOG:
+             driver = "sheepdog";
+             if (!(location = qemuBlockStorageSourceGetSheepdogProps(src)))
+diff --git a/src/qemu/qemu_domain.c b/src/qemu/qemu_domain.c
+index 953808fcfe..62860283d8 100644
+--- a/src/qemu/qemu_domain.c
++++ b/src/qemu/qemu_domain.c
+@@ -5215,7 +5215,8 @@ qemuDomainValidateStorageSource(virStorageSource *src,
+     if (src->query &&
+         (actualType != VIR_STORAGE_TYPE_NETWORK ||
+          (src->protocol != VIR_STORAGE_NET_PROTOCOL_HTTPS &&
+-          src->protocol != VIR_STORAGE_NET_PROTOCOL_HTTP))) {
++          src->protocol != VIR_STORAGE_NET_PROTOCOL_HTTP &&
++          src->protocol != VIR_STORAGE_NET_PROTOCOL_VITASTOR))) {
+         virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
+                        _("query is supported only with HTTP(S) protocols"));
+         return -1;
+@@ -10340,6 +10341,7 @@ qemuDomainPrepareStorageSourceTLS(virStorageSource *src,
+         break;
+ 
+     case VIR_STORAGE_NET_PROTOCOL_RBD:
++    case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
+     case VIR_STORAGE_NET_PROTOCOL_SHEEPDOG:
+     case VIR_STORAGE_NET_PROTOCOL_GLUSTER:
+     case VIR_STORAGE_NET_PROTOCOL_ISCSI:
+diff --git a/src/qemu/qemu_snapshot.c b/src/qemu/qemu_snapshot.c
+index 73ff533827..e9c799ca8f 100644
+--- a/src/qemu/qemu_snapshot.c
++++ b/src/qemu/qemu_snapshot.c
+@@ -423,6 +423,7 @@ qemuSnapshotPrepareDiskExternalInactive(virDomainSnapshotDiskDef *snapdisk,
+         case VIR_STORAGE_NET_PROTOCOL_NONE:
+         case VIR_STORAGE_NET_PROTOCOL_NBD:
+         case VIR_STORAGE_NET_PROTOCOL_RBD:
++        case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
+         case VIR_STORAGE_NET_PROTOCOL_SHEEPDOG:
+         case VIR_STORAGE_NET_PROTOCOL_GLUSTER:
+         case VIR_STORAGE_NET_PROTOCOL_ISCSI:
+@@ -648,6 +649,7 @@ qemuSnapshotPrepareDiskInternal(virDomainDiskDef *disk,
+         case VIR_STORAGE_NET_PROTOCOL_NONE:
+         case VIR_STORAGE_NET_PROTOCOL_NBD:
+         case VIR_STORAGE_NET_PROTOCOL_RBD:
++        case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
+         case VIR_STORAGE_NET_PROTOCOL_SHEEPDOG:
+         case VIR_STORAGE_NET_PROTOCOL_GLUSTER:
+         case VIR_STORAGE_NET_PROTOCOL_ISCSI:
+diff --git a/src/storage/storage_driver.c b/src/storage/storage_driver.c
+index 314fe930e0..fb615a8b4e 100644
+--- a/src/storage/storage_driver.c
++++ b/src/storage/storage_driver.c
+@@ -1626,6 +1626,7 @@ storageVolLookupByPathCallback(virStoragePoolObj *obj,
+ 
+         case VIR_STORAGE_POOL_GLUSTER:
+         case VIR_STORAGE_POOL_RBD:
++        case VIR_STORAGE_POOL_VITASTOR:
+         case VIR_STORAGE_POOL_SHEEPDOG:
+         case VIR_STORAGE_POOL_ZFS:
+         case VIR_STORAGE_POOL_LAST:
+diff --git a/src/storage_file/storage_source_backingstore.c b/src/storage_file/storage_source_backingstore.c
+index 80681924ea..8a3ade9ec0 100644
+--- a/src/storage_file/storage_source_backingstore.c
++++ b/src/storage_file/storage_source_backingstore.c
+@@ -287,6 +287,75 @@ virStorageSourceParseRBDColonString(const char *rbdstr,
+ }
+ 
+ 
++static int
++virStorageSourceParseVitastorColonString(const char *colonstr,
++                                         virStorageSource *src)
++{
++    char *p, *e, *next;
++    g_autofree char *options = NULL;
++
++    /* optionally skip the "vitastor:" prefix if provided */
++    if (STRPREFIX(colonstr, "vitastor:"))
++        colonstr += strlen("vitastor:");
++
++    options = g_strdup(colonstr);
++
++    p = options;
++    while (*p) {
++        /* find : delimiter or end of string */
++        for (e = p; *e && *e != ':'; ++e) {
++            if (*e == '\\') {
++                e++;
++                if (*e == '\0')
++                    break;
++            }
++        }
++        if (*e == '\0') {
++            next = e;    /* last kv pair */
++        } else {
++            next = e + 1;
++            *e = '\0';
++        }
++
++        if (STRPREFIX(p, "image=")) {
++            src->path = g_strdup(p + strlen("image="));
++        } else if (STRPREFIX(p, "etcd-prefix=")) {
++            src->query = g_strdup(p + strlen("etcd-prefix="));
++        } else if (STRPREFIX(p, "config-path=")) {
++            src->configFile = g_strdup(p + strlen("config-path="));
++        } else if (STRPREFIX(p, "etcd-host=")) {
++            char *h, *sep;
++
++            h = p + strlen("etcd-host=");
++            while (h < e) {
++                for (sep = h; sep < e; ++sep) {
++                    if (*sep == '\\' && (sep[1] == ',' ||
++                                         sep[1] == ';' ||
++                                         sep[1] == ' ')) {
++                        *sep = '\0';
++                        sep += 2;
++                        break;
++                    }
++                }
++
++                if (virStorageSourceRBDAddHost(src, h) < 0)
++                    return -1;
++
++                h = sep;
++            }
++        }
++
++        p = next;
++    }
++
++    if (!src->path) {
++        return -1;
++    }
++
++    return 0;
++}
++
++
+ static int
+ virStorageSourceParseNBDColonString(const char *nbdstr,
+                                     virStorageSource *src)
+@@ -399,6 +468,11 @@ virStorageSourceParseBackingColon(virStorageSource *src,
+             return -1;
+         break;
+ 
++    case VIR_STORAGE_NET_PROTOCOL_VITASTOR:
++        if (virStorageSourceParseVitastorColonString(path, src) < 0)
++            return -1;
++        break;
++
+     case VIR_STORAGE_NET_PROTOCOL_SHEEPDOG:
+     case VIR_STORAGE_NET_PROTOCOL_LAST:
+     case VIR_STORAGE_NET_PROTOCOL_NONE:
+@@ -975,6 +1049,54 @@ virStorageSourceParseBackingJSONRBD(virStorageSource *src,
+     return 0;
+ }
+ 
++static int
++virStorageSourceParseBackingJSONVitastor(virStorageSource *src,
++                                         virJSONValue *json,
++                                         const char *jsonstr G_GNUC_UNUSED,
++                                         int opaque G_GNUC_UNUSED)
++{
++    const char *filename;
++    const char *image = virJSONValueObjectGetString(json, "image");
++    const char *conf = virJSONValueObjectGetString(json, "config-path");
++    const char *etcd_prefix = virJSONValueObjectGetString(json, "etcd-prefix");
++    virJSONValue *servers = virJSONValueObjectGetArray(json, "server");
++    size_t nservers;
++    size_t i;
++
++    src->type = VIR_STORAGE_TYPE_NETWORK;
++    src->protocol = VIR_STORAGE_NET_PROTOCOL_VITASTOR;
++
++    /* legacy syntax passed via 'filename' option */
++    if ((filename = virJSONValueObjectGetString(json, "filename")))
++        return virStorageSourceParseVitastorColonString(filename, src);
++
++    if (!image) {
++        virReportError(VIR_ERR_INVALID_ARG, "%s",
++                       _("missing image name in Vitastor backing volume "
++                         "JSON specification"));
++        return -1;
++    }
++
++    src->path = g_strdup(image);
++    src->configFile = g_strdup(conf);
++    src->query = g_strdup(etcd_prefix);
++
++    if (servers) {
++        nservers = virJSONValueArraySize(servers);
++
++        src->hosts = g_new0(virStorageNetHostDef, nservers);
++        src->nhosts = nservers;
++
++        for (i = 0; i < nservers; i++) {
++            if (virStorageSourceParseBackingJSONInetSocketAddress(src->hosts + i,
++                                                                  virJSONValueArrayGet(servers, i)) < 0)
++                return -1;
++        }
++    }
++
++    return 0;
++}
++
+ static int
+ virStorageSourceParseBackingJSONRaw(virStorageSource *src,
+                                     virJSONValue *json,
+@@ -1152,6 +1274,7 @@ static const struct virStorageSourceJSONDriverParser jsonParsers[] = {
+     {"sheepdog", false, virStorageSourceParseBackingJSONSheepdog, 0},
+     {"ssh", false, virStorageSourceParseBackingJSONSSH, 0},
+     {"rbd", false, virStorageSourceParseBackingJSONRBD, 0},
++    {"vitastor", false, virStorageSourceParseBackingJSONVitastor, 0},
+     {"raw", true, virStorageSourceParseBackingJSONRaw, 0},
+     {"nfs", false, virStorageSourceParseBackingJSONNFS, 0},
+     {"vxhs", false, virStorageSourceParseBackingJSONVxHS, 0},
+diff --git a/src/test/test_driver.c b/src/test/test_driver.c
+index e87d7cfd44..ccc05d7aae 100644
+--- a/src/test/test_driver.c
++++ b/src/test/test_driver.c
+@@ -7335,6 +7335,7 @@ testStorageVolumeTypeForPool(int pooltype)
+     case VIR_STORAGE_POOL_ISCSI_DIRECT:
+     case VIR_STORAGE_POOL_GLUSTER:
+     case VIR_STORAGE_POOL_RBD:
++    case VIR_STORAGE_POOL_VITASTOR:
+         return VIR_STORAGE_VOL_NETWORK;
+     case VIR_STORAGE_POOL_LOGICAL:
+     case VIR_STORAGE_POOL_DISK:
+diff --git a/tests/storagepoolcapsschemadata/poolcaps-fs.xml b/tests/storagepoolcapsschemadata/poolcaps-fs.xml
+index eee75af746..8bd0a57bdd 100644
+--- a/tests/storagepoolcapsschemadata/poolcaps-fs.xml
++++ b/tests/storagepoolcapsschemadata/poolcaps-fs.xml
+@@ -204,4 +204,11 @@
+       </enum>
+     </volOptions>
+   </pool>
++  <pool type='vitastor' supported='no'>
++    <volOptions>
++      <defaultFormat type='raw'/>
++      <enum name='targetFormatType'>
++      </enum>
++    </volOptions>
++  </pool>
+ </storagepoolCapabilities>
+diff --git a/tests/storagepoolcapsschemadata/poolcaps-full.xml b/tests/storagepoolcapsschemadata/poolcaps-full.xml
+index 805950a937..852df0de16 100644
+--- a/tests/storagepoolcapsschemadata/poolcaps-full.xml
++++ b/tests/storagepoolcapsschemadata/poolcaps-full.xml
+@@ -204,4 +204,11 @@
+       </enum>
+     </volOptions>
+   </pool>
++  <pool type='vitastor' supported='yes'>
++    <volOptions>
++      <defaultFormat type='raw'/>
++      <enum name='targetFormatType'>
++      </enum>
++    </volOptions>
++  </pool>
+ </storagepoolCapabilities>
+diff --git a/tests/storagepoolxml2argvtest.c b/tests/storagepoolxml2argvtest.c
+index e8e40d695e..db55fe5f3a 100644
+--- a/tests/storagepoolxml2argvtest.c
++++ b/tests/storagepoolxml2argvtest.c
+@@ -65,6 +65,7 @@ testCompareXMLToArgvFiles(bool shouldFail,
+     case VIR_STORAGE_POOL_GLUSTER:
+     case VIR_STORAGE_POOL_ZFS:
+     case VIR_STORAGE_POOL_VSTORAGE:
++    case VIR_STORAGE_POOL_VITASTOR:
+     case VIR_STORAGE_POOL_LAST:
+     default:
+         VIR_TEST_DEBUG("pool type '%s' has no xml2argv test", defTypeStr);
+diff --git a/tools/virsh-pool.c b/tools/virsh-pool.c
+index 36f00cf643..5f5bd3464e 100644
+--- a/tools/virsh-pool.c
++++ b/tools/virsh-pool.c
+@@ -1223,6 +1223,9 @@ cmdPoolList(vshControl *ctl, const vshCmd *cmd G_GNUC_UNUSED)
+             case VIR_STORAGE_POOL_VSTORAGE:
+                 flags |= VIR_CONNECT_LIST_STORAGE_POOLS_VSTORAGE;
+                 break;
++            case VIR_STORAGE_POOL_VITASTOR:
++                flags |= VIR_CONNECT_LIST_STORAGE_POOLS_VITASTOR;
++                break;
+             case VIR_STORAGE_POOL_LAST:
+                 break;
+             }

pull_request_template.yml

@@ -0,0 +1,28 @@
+name: Pull Request
+about: Submit a pull request
+body:
+  - type: textarea
+    id: description
+    attributes:
+      label: Description
+      description: Describe your pull request
+      placeholder: ""
+      value: ""
+    validations:
+      required: true
+  - type: input
+    id: author
+    attributes:
+      label: Contributor Name
+      description: Contributor Name or Company Details if the Contributor is a company
+      placeholder: ""
+    validations:
+      required: false
+  - type: checkboxes
+    id: terms
+    attributes:
+      label: CLA
+      description: By submitting this pull request, I accept [Vitastor CLA](https://git.yourcmc.ru/vitalif/vitastor/src/branch/master/CLA-en.md)
+      options:
+        - label: "I accept Vitastor CLA agreement: https://git.yourcmc.ru/vitalif/vitastor/src/branch/master/CLA-en.md"
+          required: true

rpm/build-tarball.sh

@@ -24,4 +24,4 @@ rm fio
 mv fio-copy fio
 FIO=`rpm -qi fio | perl -e 'while(<>) { /^Epoch[\s:]+(\S+)/ && print "$1:"; /^Version[\s:]+(\S+)/ && print $1; /^Release[\s:]+(\S+)/ && print "-$1"; }'`
 perl -i -pe 's/(Requires:\s*fio)([^\n]+)?/$1 = '$FIO'/' $VITASTOR/rpm/vitastor-el$EL.spec
-tar --transform 's#^#vitastor-1.3.1/#' --exclude 'rpm/*.rpm' -czf $VITASTOR/../vitastor-1.3.1$(rpm --eval '%dist').tar.gz *
+tar --transform 's#^#vitastor-1.4.5/#' --exclude 'rpm/*.rpm' -czf $VITASTOR/../vitastor-1.4.5$(rpm --eval '%dist').tar.gz *

rpm/vitastor-el7.Dockerfile

@@ -36,7 +36,7 @@ ADD . /root/vitastor
 RUN set -e; \
     cd /root/vitastor/rpm; \
     sh build-tarball.sh; \
-    cp /root/vitastor-1.3.1.el7.tar.gz ~/rpmbuild/SOURCES; \
+    cp /root/vitastor-1.4.5.el7.tar.gz ~/rpmbuild/SOURCES; \
     cp vitastor-el7.spec ~/rpmbuild/SPECS/vitastor.spec; \
     cd ~/rpmbuild/SPECS/; \
     rpmbuild -ba vitastor.spec; \

rpm/vitastor-el7.spec

@@ -1,11 +1,11 @@
 Name:           vitastor
-Version:        1.3.1
+Version:        1.4.5
 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-1.3.1.el7.tar.gz
+Source0:        vitastor-1.4.5.el7.tar.gz
 
 BuildRequires:  liburing-devel >= 0.6
 BuildRequires:  gperftools-devel

rpm/vitastor-el8.Dockerfile

@@ -35,7 +35,7 @@ ADD . /root/vitastor
 RUN set -e; \
     cd /root/vitastor/rpm; \
     sh build-tarball.sh; \
-    cp /root/vitastor-1.3.1.el8.tar.gz ~/rpmbuild/SOURCES; \
+    cp /root/vitastor-1.4.5.el8.tar.gz ~/rpmbuild/SOURCES; \
     cp vitastor-el8.spec ~/rpmbuild/SPECS/vitastor.spec; \
     cd ~/rpmbuild/SPECS/; \
     rpmbuild -ba vitastor.spec; \

rpm/vitastor-el8.spec

@@ -1,11 +1,11 @@
 Name:           vitastor
-Version:        1.3.1
+Version:        1.4.5
 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-1.3.1.el8.tar.gz
+Source0:        vitastor-1.4.5.el8.tar.gz
 
 BuildRequires:  liburing-devel >= 0.6
 BuildRequires:  gperftools-devel

rpm/vitastor-el9.Dockerfile

@@ -18,7 +18,7 @@ ADD . /root/vitastor
 RUN set -e; \
     cd /root/vitastor/rpm; \
     sh build-tarball.sh; \
-    cp /root/vitastor-1.3.1.el9.tar.gz ~/rpmbuild/SOURCES; \
+    cp /root/vitastor-1.4.5.el9.tar.gz ~/rpmbuild/SOURCES; \
     cp vitastor-el9.spec ~/rpmbuild/SPECS/vitastor.spec; \
     cd ~/rpmbuild/SPECS/; \
     rpmbuild -ba vitastor.spec; \

rpm/vitastor-el9.spec

@@ -1,11 +1,11 @@
 Name:           vitastor
-Version:        1.3.1
+Version:        1.4.5
 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-1.3.1.el9.tar.gz
+Source0:        vitastor-1.4.5.el9.tar.gz
 
 BuildRequires:  liburing-devel >= 0.6
 BuildRequires:  gperftools-devel

src/CMakeLists.txt

@@ -16,7 +16,7 @@ if("${CMAKE_INSTALL_PREFIX}" MATCHES "^/usr/local/?$")
 	set(CMAKE_INSTALL_RPATH "${CMAKE_INSTALL_PREFIX}/${CMAKE_INSTALL_LIBDIR}")
 endif()
 
-add_definitions(-DVERSION="1.3.1")
+add_definitions(-DVERSION="1.4.5")
 add_definitions(-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})

src/blockstore_flush.cpp

@@ -19,7 +19,6 @@ journal_flusher_t::journal_flusher_t(blockstore_impl_t *bs)
     syncing_flushers = 0;
     // FIXME: allow to configure flusher_start_threshold and journal_trim_interval
     flusher_start_threshold = bs->dsk.journal_block_size / sizeof(journal_entry_stable);
-    journal_trim_interval = 512;
     journal_trim_counter = bs->journal.flush_journal ? 1 : 0;
     trim_wanted = bs->journal.flush_journal ? 1 : 0;
     journal_superblock = bs->journal.inmemory ? bs->journal.buffer : memalign_or_die(MEM_ALIGNMENT, bs->dsk.journal_block_size);
@@ -184,8 +183,7 @@ void journal_flusher_t::mark_trim_possible()
     if (trim_wanted > 0)
     {
         dequeuing = true;
-        if (!journal_trim_counter)
-            journal_trim_counter = journal_trim_interval;
+        journal_trim_counter = 0;
         bs->ringloop->wakeup();
     }
 }
@@ -366,9 +364,10 @@ resume_0:
         !flusher->flush_queue.size() || !flusher->dequeuing)
     {
 stop_flusher:
-        if (flusher->trim_wanted > 0 && flusher->journal_trim_counter > 0)
+        if (flusher->trim_wanted > 0 && cur.oid.inode != 0)
         {
             // Attempt forced trim
+            cur.oid = {};
             flusher->active_flushers++;
             goto trim_journal;
         }
@@ -584,7 +583,8 @@ resume_2:
         flusher->sync_to_repeat.erase(repeat_it);
     trim_journal:
         // Clear unused part of the journal every <journal_trim_interval> flushes
-        if (!((++flusher->journal_trim_counter) % flusher->journal_trim_interval) || flusher->trim_wanted > 0)
+        if (bs->journal_trim_interval && !((++flusher->journal_trim_counter) % bs->journal_trim_interval) ||
+            flusher->trim_wanted > 0)
         {
     resume_26:
     resume_27:
@@ -1346,7 +1346,6 @@ bool journal_flusher_co::trim_journal(int wait_base)
     else if (wait_state == wait_base+2) goto resume_2;
     else if (wait_state == wait_base+3) goto resume_3;
     else if (wait_state == wait_base+4) goto resume_4;
-    flusher->journal_trim_counter = 0;
     new_trim_pos = bs->journal.get_trim_pos();
     if (new_trim_pos != bs->journal.used_start)
     {
@@ -1419,6 +1418,7 @@ bool journal_flusher_co::trim_journal(int wait_base)
                 exit(0);
             }
         }
+        flusher->journal_trim_counter = 0;
         flusher->trimming = false;
     }
     return true;

src/blockstore_flush.h

@@ -107,7 +107,7 @@ class journal_flusher_t
     blockstore_impl_t *bs;
     friend class journal_flusher_co;
 
-    int journal_trim_counter, journal_trim_interval;
+    int journal_trim_counter;
     bool trimming;
     void* journal_superblock;
 

src/blockstore_impl.cpp

@@ -163,20 +163,10 @@ void blockstore_impl_t::loop()
             }
             else if (op->opcode == BS_OP_SYNC)
             {
-                // wait for all small writes to be submitted
-                // wait for all big writes to complete, submit data device fsync
+                // sync only completed writes?
                 // wait for the data device fsync to complete, then submit journal writes for big writes
                 // then submit an fsync operation
-                if (has_writes)
-                {
-                    // Can't submit SYNC before previous writes
-                    continue;
-                }
                 wr_st = continue_sync(op);
-                if (wr_st != 2)
-                {
-                    has_writes = wr_st > 0 ? 1 : 2;
-                }
             }
             else if (op->opcode == BS_OP_STABLE)
             {
@@ -205,6 +195,10 @@ void blockstore_impl_t::loop()
                     // ring is full, stop submission
                     break;
                 }
+                else if (PRIV(op)->wait_for == WAIT_JOURNAL)
+                {
+                    PRIV(op)->wait_detail2 = (unstable_writes.size()+unstable_unsynced);
+                }
             }
         }
         if (op_idx != new_idx)
@@ -283,7 +277,8 @@ void blockstore_impl_t::check_wait(blockstore_op_t *op)
     }
     else if (PRIV(op)->wait_for == WAIT_JOURNAL)
     {
-        if (journal.used_start == PRIV(op)->wait_detail)
+        if (journal.used_start == PRIV(op)->wait_detail &&
+            (unstable_writes.size()+unstable_unsynced) == PRIV(op)->wait_detail2)
         {
             // do not submit
 #ifdef BLOCKSTORE_DEBUG
@@ -558,13 +553,14 @@ void blockstore_impl_t::process_list(blockstore_op_t *op)
             if (stable_count >= stable_alloc)
             {
                 stable_alloc *= 2;
-                stable = (obj_ver_id*)realloc(stable, sizeof(obj_ver_id) * stable_alloc);
-                if (!stable)
+                obj_ver_id* nst = (obj_ver_id*)realloc(stable, sizeof(obj_ver_id) * stable_alloc);
+                if (!nst)
                 {
                     op->retval = -ENOMEM;
                     FINISH_OP(op);
                     return;
                 }
+                stable = nst;
             }
             stable[stable_count++] = {
                 .oid = clean_it->first,
@@ -642,8 +638,8 @@ void blockstore_impl_t::process_list(blockstore_op_t *op)
                             if (stable_count >= stable_alloc)
                             {
                                 stable_alloc += 32768;
-                                stable = (obj_ver_id*)realloc(stable, sizeof(obj_ver_id) * stable_alloc);
-                                if (!stable)
+                                obj_ver_id *nst = (obj_ver_id*)realloc(stable, sizeof(obj_ver_id) * stable_alloc);
+                                if (!nst)
                                 {
                                     if (unstable)
                                         free(unstable);
@@ -651,6 +647,7 @@ void blockstore_impl_t::process_list(blockstore_op_t *op)
                                     FINISH_OP(op);
                                     return;
                                 }
+                                stable = nst;
                             }
                             stable[stable_count++] = dirty_it->first;
                         }
@@ -666,8 +663,8 @@ void blockstore_impl_t::process_list(blockstore_op_t *op)
                     if (unstable_count >= unstable_alloc)
                     {
                         unstable_alloc += 32768;
-                        unstable = (obj_ver_id*)realloc(unstable, sizeof(obj_ver_id) * unstable_alloc);
-                        if (!unstable)
+                        obj_ver_id *nst = (obj_ver_id*)realloc(unstable, sizeof(obj_ver_id) * unstable_alloc);
+                        if (!nst)
                         {
                             if (stable)
                                 free(stable);
@@ -675,6 +672,7 @@ void blockstore_impl_t::process_list(blockstore_op_t *op)
                             FINISH_OP(op);
                             return;
                         }
+                        unstable = nst;
                     }
                     unstable[unstable_count++] = dirty_it->first;
                 }
@@ -694,8 +692,8 @@ void blockstore_impl_t::process_list(blockstore_op_t *op)
     if (stable_count+unstable_count > stable_alloc)
     {
         stable_alloc = stable_count+unstable_count;
-        stable = (obj_ver_id*)realloc(stable, sizeof(obj_ver_id) * stable_alloc);
-        if (!stable)
+        obj_ver_id *nst = (obj_ver_id*)realloc(stable, sizeof(obj_ver_id) * stable_alloc);
+        if (!nst)
         {
             if (unstable)
                 free(unstable);
@@ -703,6 +701,7 @@ void blockstore_impl_t::process_list(blockstore_op_t *op)
             FINISH_OP(op);
             return;
         }
+        stable = nst;
     }
     // Copy unstable entries
     for (int i = 0; i < unstable_count; i++)

src/blockstore_impl.h

@@ -55,6 +55,7 @@
 #define IS_JOURNAL(st) (((st) & 0x0F) == BS_ST_SMALL_WRITE)
 #define IS_BIG_WRITE(st) (((st) & 0x0F) == BS_ST_BIG_WRITE)
 #define IS_DELETE(st) (((st) & 0x0F) == BS_ST_DELETE)
+#define IS_INSTANT(st) (((st) & BS_ST_TYPE_MASK) == BS_ST_DELETE || ((st) & BS_ST_INSTANT))
 
 #define BS_SUBMIT_CHECK_SQES(n) \
     if (ringloop->sqes_left() < (n))\
@@ -201,7 +202,7 @@ struct blockstore_op_private_t
 {
     // Wait status
     int wait_for;
-    uint64_t wait_detail;
+    uint64_t wait_detail, wait_detail2;
     int pending_ops;
     int op_state;
 
@@ -252,6 +253,7 @@ class blockstore_impl_t
     bool inmemory_meta = false;
     // Maximum and minimum flusher count
     unsigned max_flusher_count, min_flusher_count;
+    unsigned journal_trim_interval;
     // Maximum queue depth
     unsigned max_write_iodepth = 128;
     // Enable small (journaled) write throttling, useful for the SSD+HDD case
@@ -377,7 +379,7 @@ class blockstore_impl_t
     // Stabilize
     int dequeue_stable(blockstore_op_t *op);
     int continue_stable(blockstore_op_t *op);
-    void mark_stable(const obj_ver_id & ov, bool forget_dirty = false);
+    void mark_stable(obj_ver_id ov, bool forget_dirty = false);
     void stabilize_object(object_id oid, uint64_t max_ver);
     blockstore_op_t* selective_sync(blockstore_op_t *op);
     int split_stab_op(blockstore_op_t *op, std::function<int(obj_ver_id v)> decider);

src/blockstore_open.cpp

@@ -13,13 +13,14 @@ void blockstore_impl_t::parse_config(blockstore_config_t & config, bool init)
         max_flusher_count = strtoull(config["flusher_count"].c_str(), NULL, 10);
     }
     min_flusher_count = strtoull(config["min_flusher_count"].c_str(), NULL, 10);
+    journal_trim_interval = strtoull(config["journal_trim_interval"].c_str(), NULL, 10);
     max_write_iodepth = strtoull(config["max_write_iodepth"].c_str(), NULL, 10);
     throttle_small_writes = config["throttle_small_writes"] == "true" || config["throttle_small_writes"] == "1" || config["throttle_small_writes"] == "yes";
     throttle_target_iops = strtoull(config["throttle_target_iops"].c_str(), NULL, 10);
     throttle_target_mbs = strtoull(config["throttle_target_mbs"].c_str(), NULL, 10);
     throttle_target_parallelism = strtoull(config["throttle_target_parallelism"].c_str(), NULL, 10);
     throttle_threshold_us = strtoull(config["throttle_threshold_us"].c_str(), NULL, 10);
-    if (config.find("autosync_writes") != config.end())
+    if (config["autosync_writes"] != "")
     {
         autosync_writes = strtoull(config["autosync_writes"].c_str(), NULL, 10);
     }
@@ -31,6 +32,10 @@ void blockstore_impl_t::parse_config(blockstore_config_t & config, bool init)
     {
         min_flusher_count = 1;
     }
+    if (!journal_trim_interval)
+    {
+        journal_trim_interval = 512;
+    }
     if (!max_write_iodepth)
     {
         max_write_iodepth = 128;

src/blockstore_stable.cpp

@@ -412,11 +412,40 @@ resume_4:
     return 2;
 }
 
-void blockstore_impl_t::mark_stable(const obj_ver_id & v, bool forget_dirty)
+void blockstore_impl_t::mark_stable(obj_ver_id v, bool forget_dirty)
 {
     auto dirty_it = dirty_db.find(v);
     if (dirty_it != dirty_db.end())
     {
+        if (IS_INSTANT(dirty_it->second.state))
+        {
+            // 'Instant' (non-EC) operations may complete and try to become stable out of order. Prevent it.
+            auto back_it = dirty_it;
+            while (back_it != dirty_db.begin())
+            {
+                back_it--;
+                if (back_it->first.oid != v.oid)
+                {
+                    break;
+                }
+                if (!IS_STABLE(back_it->second.state))
+                {
+                    // There are preceding unstable versions, can't flush <v>
+                    return;
+                }
+            }
+            while (true)
+            {
+                dirty_it++;
+                if (dirty_it == dirty_db.end() || dirty_it->first.oid != v.oid ||
+                    !IS_SYNCED(dirty_it->second.state))
+                {
+                    dirty_it--;
+                    break;
+                }
+                v.version = dirty_it->first.version;
+            }
+        }
         while (1)
         {
             bool was_stable = IS_STABLE(dirty_it->second.state);

src/blockstore_sync.cpp

@@ -76,6 +76,7 @@ int blockstore_impl_t::continue_sync(blockstore_op_t *op)
         // 2nd step: Data device is synced, prepare & write journal entries
         // Check space in the journal and journal memory buffers
         blockstore_journal_check_t space_check(this);
+        auto reservation = (unstable_writes.size()+unstable_unsynced+PRIV(op)->sync_big_writes.size())*journal.block_size;
         if (dsk.csum_block_size)
         {
             // More complex check because all journal entries have different lengths
@@ -85,16 +86,14 @@ int blockstore_impl_t::continue_sync(blockstore_op_t *op)
                 left--;
                 auto & dirty_entry = dirty_db.at(sbw);
                 uint64_t dyn_size = dsk.dirty_dyn_size(dirty_entry.offset, dirty_entry.len);
-                if (!space_check.check_available(op, 1, sizeof(journal_entry_big_write) + dyn_size,
-                    (unstable_writes.size()+unstable_unsynced)*journal.block_size))
+                if (!space_check.check_available(op, 1, sizeof(journal_entry_big_write) + dyn_size, left ? 0 : reservation))
                 {
                     return 0;
                 }
             }
         }
         else if (!space_check.check_available(op, PRIV(op)->sync_big_writes.size(),
-            sizeof(journal_entry_big_write) + dsk.clean_entry_bitmap_size,
-            (unstable_writes.size()+unstable_unsynced)*journal.block_size))
+            sizeof(journal_entry_big_write) + dsk.clean_entry_bitmap_size, reservation))
         {
             return 0;
         }

src/blockstore_write.cpp

@@ -129,7 +129,7 @@ bool blockstore_impl_t::enqueue_write(blockstore_op_t *op)
     }
     bool imm = (op->len < dsk.data_block_size ? (immediate_commit != IMMEDIATE_NONE) : (immediate_commit == IMMEDIATE_ALL));
     if (wait_big && !is_del && !deleted && op->len < dsk.data_block_size && !imm ||
-        !imm && unsynced_queued_ops >= autosync_writes)
+        !imm && autosync_writes && unsynced_queued_ops >= autosync_writes)
     {
         // Issue an additional sync so that the previous big write can reach the journal
         blockstore_op_t *sync_op = new blockstore_op_t;
@@ -320,7 +320,7 @@ int blockstore_impl_t::dequeue_write(blockstore_op_t *op)
         blockstore_journal_check_t space_check(this);
         if (!space_check.check_available(op, unsynced_big_write_count + 1,
             sizeof(journal_entry_big_write) + dsk.clean_dyn_size,
-            (unstable_writes.size()+unstable_unsynced)*journal.block_size))
+            (unstable_writes.size()+unstable_unsynced+((dirty_it->second.state & BS_ST_INSTANT) ? 0 : 1))*journal.block_size))
         {
             return 0;
         }
@@ -412,7 +412,7 @@ int blockstore_impl_t::dequeue_write(blockstore_op_t *op)
                 sizeof(journal_entry_big_write) + dsk.clean_dyn_size, 0)
             || !space_check.check_available(op, 1,
                 sizeof(journal_entry_small_write) + dyn_size,
-                op->len + (unstable_writes.size()+unstable_unsynced)*journal.block_size))
+                op->len + (unstable_writes.size()+unstable_unsynced+((dirty_it->second.state & BS_ST_INSTANT) ? 0 : 1))*journal.block_size))
         {
             return 0;
         }
@@ -549,7 +549,7 @@ resume_2:
         uint64_t dyn_size = dsk.dirty_dyn_size(op->offset, op->len);
         blockstore_journal_check_t space_check(this);
         if (!space_check.check_available(op, 1, sizeof(journal_entry_big_write) + dyn_size,
-            (unstable_writes.size()+unstable_unsynced)*journal.block_size))
+            (unstable_writes.size()+unstable_unsynced+((dirty_it->second.state & BS_ST_INSTANT) ? 0 : 1))*journal.block_size))
         {
             return 0;
         }
@@ -593,7 +593,7 @@ resume_4:
 #endif
         bool is_big = (dirty_it->second.state & BS_ST_TYPE_MASK) == BS_ST_BIG_WRITE;
         bool imm = is_big ? (immediate_commit == IMMEDIATE_ALL) : (immediate_commit != IMMEDIATE_NONE);
-        bool is_instant = ((dirty_it->second.state & BS_ST_TYPE_MASK) == BS_ST_DELETE || (dirty_it->second.state & BS_ST_INSTANT));
+        bool is_instant = IS_INSTANT(dirty_it->second.state);
         if (imm)
         {
             auto & unstab = unstable_writes[op->oid];

src/cli_rm_data.cpp

@@ -17,6 +17,7 @@ struct rm_pg_t
     uint64_t obj_count = 0, obj_done = 0;
     int state = 0;
     int in_flight = 0;
+    bool synced = false;
 };
 
 struct rm_inode_t
@@ -48,6 +49,7 @@ struct rm_inode_t
                 .objects = objects,
                 .obj_count = objects.size(),
                 .obj_done = 0,
+                .synced = parent->cli->get_immediate_commit(inode),
             });
             if (min_offset == 0)
             {
@@ -151,6 +153,37 @@ struct rm_inode_t
             }
             cur_list->obj_pos++;
         }
+        if (cur_list->in_flight == 0 && cur_list->obj_pos == cur_list->objects.end() &&
+            !cur_list->synced)
+        {
+            osd_op_t *op = new osd_op_t();
+            op->op_type = OSD_OP_OUT;
+            op->peer_fd = parent->cli->msgr.osd_peer_fds.at(cur_list->rm_osd_num);
+            op->req = (osd_any_op_t){
+                .sync = {
+                    .header = {
+                        .magic = SECONDARY_OSD_OP_MAGIC,
+                        .id = parent->cli->next_op_id(),
+                        .opcode = OSD_OP_SYNC,
+                    },
+                },
+            };
+            op->callback = [this, cur_list](osd_op_t *op)
+            {
+                cur_list->in_flight--;
+                cur_list->synced = true;
+                if (op->reply.hdr.retval < 0)
+                {
+                    fprintf(stderr, "Failed to sync OSD %lu (retval=%ld)\n",
+                        cur_list->rm_osd_num, op->reply.hdr.retval);
+                    error_count++;
+                }
+                delete op;
+                continue_delete();
+            };
+            cur_list->in_flight++;
+            parent->cli->msgr.outbox_push(op);
+        }
     }
 
     void continue_delete()
@@ -161,7 +194,8 @@ struct rm_inode_t
         }
         for (int i = 0; i < lists.size(); i++)
         {
-            if (!lists[i]->in_flight && lists[i]->obj_pos == lists[i]->objects.end())
+            if (!lists[i]->in_flight && lists[i]->obj_pos == lists[i]->objects.end() &&
+                lists[i]->synced)
             {
                 delete lists[i];
                 lists.erase(lists.begin()+i, lists.begin()+i+1);
@@ -187,7 +221,7 @@ struct rm_inode_t
             {
                 fprintf(stderr, "\n");
             }
-            if (parent->progress && (total_done < total_count || inactive_osds.size() > 0))
+            if (parent->progress && (total_done < total_count || inactive_osds.size() > 0 || error_count > 0))
             {
                 fprintf(
                     stderr, "Warning: Pool:%u,ID:%lu inode data may not have been fully removed.\n"

src/cli_status.cpp

@@ -106,7 +106,7 @@ resume_2:
             if (etcd_states[i]["error"].is_null())
             {
                 etcd_alive++;
-                etcd_db_size = etcd_states[i]["dbSizeInUse"].uint64_value();
+                etcd_db_size = etcd_states[i]["dbSize"].uint64_value();
             }
         }
         int mon_count = 0;

src/cluster_client.cpp

@@ -352,13 +352,13 @@ void cluster_client_t::on_load_config_hook(json11::Json::object & etcd_global_co
     // up_wait_retry_interval
     up_wait_retry_interval = config["up_wait_retry_interval"].uint64_value();
     if (!up_wait_retry_interval)
-    {
-        up_wait_retry_interval = 500;
-    }
-    else if (up_wait_retry_interval < 50)
     {
         up_wait_retry_interval = 50;
     }
+    else if (up_wait_retry_interval < 10)
+    {
+        up_wait_retry_interval = 10;
+    }
     // log_level
     log_level = config["log_level"].uint64_value();
     msgr.parse_config(config);

src/disk_tool_prepare.cpp

@@ -8,6 +8,7 @@
 int disk_tool_t::prepare_one(std::map<std::string, std::string> options, int is_hdd)
 {
     static const char *allow_additional_params[] = {
+        "autosync_writes",
         "data_io",
         "meta_io",
         "journal_io",
@@ -99,12 +100,9 @@ int disk_tool_t::prepare_one(std::map<std::string, std::string> options, int is_
         options["disable_journal_fsync"] = options["disable_data_fsync"];
     }
     // Calculate offsets if the same device is used for two or more of data, meta, and journal
-    if (options["journal_size"] == "")
+    if (options["journal_size"] == "" && (options["journal_device"] == "" || options["journal_device"] == options["data_device"]))
     {
-        if (options["journal_device"] == "")
-            options["journal_size"] = is_hdd ? "128M" : "32M";
-        else if (is_hdd)
-            options["journal_size"] = DEFAULT_HYBRID_JOURNAL;
+        options["journal_size"] = is_hdd || !json_is_true(options["disable_data_fsync"]) ? "128M" : "32M";
     }
     bool is_hybrid = is_hdd && options["journal_device"] != "" && options["journal_device"] != options["data_device"];
     if (is_hdd)
@@ -114,6 +112,15 @@ int disk_tool_t::prepare_one(std::map<std::string, std::string> options, int is_
         if (is_hybrid && options["throttle_small_writes"] == "")
             options["throttle_small_writes"] = "1";
     }
+    else if (!json_is_true(options["disable_data_fsync"]))
+    {
+        if (options.find("min_flusher_count") == options.end())
+            options["min_flusher_count"] = "32";
+        if (options.find("max_flusher_count") == options.end())
+            options["max_flusher_count"] = "256";
+        if (options.find("autosync_writes") == options.end())
+            options["autosync_writes"] = "512";
+    }
     json11::Json::object sb;
     blockstore_disk_t dsk;
     try
@@ -616,6 +623,7 @@ int disk_tool_t::prepare(std::vector<std::string> devices)
         options.erase("disable_meta_fsync");
         options.erase("disable_journal_fsync");
     }
+    auto journal_size = options["journal_size"];
     for (auto & dev: devinfo)
     {
         if (!hybrid || dev.is_hdd)
@@ -633,11 +641,13 @@ int disk_tool_t::prepare(std::vector<std::string> devices)
                     {
                         return 1;
                     }
+                    options.erase("journal_size");
                 }
                 // Treat all disks as SSDs if not in the hybrid mode
                 prepare_one(options, dev.is_hdd ? 1 : 0);
                 if (hybrid)
                 {
+                    options["journal_size"] = journal_size;
                     options.erase("journal_device");
                     options.erase("meta_device");
                 }

src/etcd_state_client.h

@@ -28,7 +28,7 @@ struct etcd_kv_t
 {
     std::string key;
     json11::Json value;
-    uint64_t mod_revision;
+    uint64_t mod_revision = 0;
 };
 
 struct pg_config_t

src/messenger.cpp

@@ -45,11 +45,12 @@ void osd_messenger_t::init()
 #endif
     keepalive_timer_id = tfd->set_timer(1000, true, [this](int)
     {
-        std::vector<int> to_stop;
-        std::vector<osd_op_t*> to_ping;
-        for (auto cl_it = clients.begin(); cl_it != clients.end(); cl_it++)
+        auto cl_it = clients.begin();
+        while (cl_it != clients.end())
         {
             auto cl = cl_it->second;
+            cl_it++;
+            auto peer_fd = cl->peer_fd;
             if (!cl->osd_num || cl->peer_state != PEER_CONNECTED && cl->peer_state != PEER_RDMA)
             {
                 // Do not run keepalive on regular clients
@@ -62,7 +63,9 @@ void osd_messenger_t::init()
                 {
                     // Ping timed out, stop the client
                     fprintf(stderr, "Ping timed out for OSD %lu (client %d), disconnecting peer\n", cl->osd_num, cl->peer_fd);
-                    to_stop.push_back(cl->peer_fd);
+                    stop_client(peer_fd, true);
+                    // Restart iterator because it may be invalidated
+                    cl_it = clients.upper_bound(peer_fd);
                 }
             }
             else if (cl->idle_time_remaining > 0)
@@ -100,9 +103,11 @@ void osd_messenger_t::init()
                             stop_client(fail_fd, true);
                         }
                     };
-                    to_ping.push_back(op);
                     cl->ping_time_remaining = osd_ping_timeout;
                     cl->idle_time_remaining = osd_idle_timeout;
+                    outbox_push(op);
+                    // Restart iterator because it may be invalidated
+                    cl_it = clients.upper_bound(peer_fd);
                 }
             }
             else
@@ -110,15 +115,6 @@ void osd_messenger_t::init()
                 cl->idle_time_remaining = osd_idle_timeout;
             }
         }
-        // Don't stop clients while a 'clients' iterator is still active
-        for (int peer_fd: to_stop)
-        {
-            stop_client(peer_fd, true);
-        }
-        for (auto op: to_ping)
-        {
-            outbox_push(op);
-        }
     });
 }
 

src/osd.cpp

@@ -22,7 +22,7 @@ static blockstore_config_t json_to_bs(const json11::Json::object & config)
     {
         if (kv.second.is_string())
             bs[kv.first] = kv.second.string_value();
-        else
+        else if (!kv.second.is_null())
             bs[kv.first] = kv.second.dump();
     }
     return bs;
@@ -194,7 +194,8 @@ void osd_t::parse_config(bool init)
         if (autosync_interval > MAX_AUTOSYNC_INTERVAL)
             autosync_interval = DEFAULT_AUTOSYNC_INTERVAL;
     }
-    if (!config["autosync_writes"].is_null())
+    if (config["autosync_writes"].is_number() ||
+        config["autosync_writes"].string_value() != "")
     {
         // Allow to set it to 0
         autosync_writes = config["autosync_writes"].uint64_value();
@@ -232,6 +233,8 @@ void osd_t::parse_config(bool init)
         ? 10 : config["recovery_tune_agg_interval"].uint64_value();
     recovery_tune_sleep_min_us = config["recovery_tune_sleep_min_us"].is_null()
         ? 10 : config["recovery_tune_sleep_min_us"].uint64_value();
+    recovery_tune_sleep_cutoff_us = config["recovery_tune_sleep_cutoff_us"].is_null()
+        ? 10000000 : config["recovery_tune_sleep_cutoff_us"].uint64_value();
     recovery_pg_switch = config["recovery_pg_switch"].uint64_value();
     if (recovery_pg_switch < 1)
         recovery_pg_switch = DEFAULT_RECOVERY_PG_SWITCH;

src/osd.h

@@ -125,6 +125,7 @@ class osd_t
     int recovery_tune_interval = 1;
     int recovery_tune_agg_interval = 10;
     int recovery_tune_sleep_min_us = 10;
+    int recovery_tune_sleep_cutoff_us = 10000000;
     int recovery_pg_switch = DEFAULT_RECOVERY_PG_SWITCH;
     int recovery_sync_batch = DEFAULT_RECOVERY_BATCH;
     int inode_vanish_time = 60;
@@ -282,6 +283,7 @@ class osd_t
     void exec_sync_stab_all(osd_op_t *cur_op);
     void exec_show_config(osd_op_t *cur_op);
     void exec_secondary(osd_op_t *cur_op);
+    void exec_secondary_real(osd_op_t *cur_op);
     void secondary_op_callback(osd_op_t *cur_op);
 
     // primary ops

src/osd_cluster.cpp

@@ -262,7 +262,8 @@ void osd_t::report_statistics()
     for (auto st_it = inode_stats.begin(); st_it != inode_stats.end(); )
     {
         auto & kv = *st_it;
-        if (!bs_inode_space[kv.first])
+        auto spc_it = bs_inode_space.find(kv.first);
+        if (spc_it == bs_inode_space.end() || !spc_it->second) // prevent autovivification
         {
             // Is it an empty inode?
             if (!tv_now.tv_sec)
@@ -842,7 +843,13 @@ void osd_t::report_pg_states()
                     pg_state_exists = true;
                     if (pg.state == PG_OFFLINE && pg_it->second.cur_primary != this->osd_num)
                     {
-                        // Nothing to check or report, PG is already taken over by another OSD
+                        // Nothing to report, PG is already taken over by another OSD
+                        checks.push_back(json11::Json::object {
+                            { "target", "MOD" },
+                            { "key", state_key_base64 },
+                            { "result", "LESS" },
+                            { "mod_revision", st_cli.etcd_watch_revision+1 },
+                        });
                         continue;
                     }
                 }
@@ -850,11 +857,6 @@ void osd_t::report_pg_states()
         }
         if (!pg_state_exists)
         {
-            if (pg.state == PG_OFFLINE)
-            {
-                // Nothing to check or report, PG is already stopped
-                continue;
-            }
             // Check that the PG key does not exist
             // Failed check indicates an unsuccessful PG lock attempt in this case
             checks.push_back(json11::Json::object {

src/osd_flush.cpp

@@ -422,6 +422,10 @@ void osd_t::tune_recovery()
     rtune_avg_lat = total_recovery_usec/recovery_count;
     uint64_t target_lat = rtune_avg_lat * rtune_avg_lat/1000000.0 * recovery_count/recovery_tune_interval / rtune_target_util;
     auto sleep_us = target_lat > rtune_avg_lat+recovery_tune_sleep_min_us ? target_lat-rtune_avg_lat : 0;
+    if (sleep_us > recovery_tune_sleep_cutoff_us)
+    {
+        return;
+    }
     if (recovery_target_sleep_items.size() != recovery_tune_agg_interval)
     {
         recovery_target_sleep_items.resize(recovery_tune_agg_interval);
@@ -438,7 +442,7 @@ void osd_t::tune_recovery()
     if (recovery_target_sleep_count < recovery_tune_agg_interval)
         recovery_target_sleep_count++;
     recovery_target_sleep_us = recovery_target_sleep_total / recovery_target_sleep_count;
-    if (log_level > 4)
+    if (log_level > 1)
     {
         printf(
             "[OSD %lu] auto-tune: client util: %.2f, recovery util: %.2f, lat: %lu us -> target util %.2f, delay %lu us\n",

src/osd_peering.cpp

@@ -222,6 +222,9 @@ void osd_t::start_pg_peering(pg_t & pg)
     }
     if (pg.pg_cursize < pg.pg_minsize)
     {
+        // FIXME: Incomplete EC PGs may currently easily lead to write hangs ("slow ops" in OSD logs)
+        // because such PGs don't flush unstable entries on secondary OSDs so they can't remove these
+        // entries from their journals...
         pg.state = PG_INCOMPLETE;
         report_pg_state(pg);
         return;

src/osd_primary.cpp

@@ -706,6 +706,26 @@ resume_5:
         remove_object_from_state(op_data->oid, &op_data->object_state, pg);
         deref_object_state(pg, &op_data->object_state, true);
     }
+    // Mark PG and OSDs as dirty
+    for (auto & chunk: (op_data->object_state ? op_data->object_state->osd_set : pg.cur_loc_set))
+    {
+        this->dirty_osds.insert(chunk.osd_num);
+    }
+    for (auto cl_it = msgr.clients.find(cur_op->peer_fd); cl_it != msgr.clients.end(); )
+    {
+        cl_it->second->dirty_pgs.insert({ .pool_id = pg.pool_id, .pg_num = pg.pg_num });
+        break;
+    }
+    dirty_pgs.insert({ .pool_id = pg.pool_id, .pg_num = pg.pg_num });
+    if (immediate_commit == IMMEDIATE_NONE)
+    {
+        unstable_write_count++;
+        if (unstable_write_count >= autosync_writes)
+        {
+            unstable_write_count = 0;
+            autosync();
+        }
+    }
     pg.total_count--;
     cur_op->reply.hdr.retval = 0;
 continue_others:

src/osd_rmw.cpp

@@ -861,15 +861,15 @@ static void calc_rmw_parity_copy_mod(osd_rmw_stripe_t *stripes, int pg_size, int
 static void calc_rmw_parity_copy_parity(osd_rmw_stripe_t *stripes, int pg_size, int pg_minsize,
     uint64_t *read_osd_set, uint64_t *write_osd_set, uint32_t chunk_size, uint32_t start, uint32_t end)
 {
-    if (write_osd_set != read_osd_set)
+    if (write_osd_set != read_osd_set && end != 0)
     {
         for (int role = pg_minsize; role < pg_size; role++)
         {
-            if (write_osd_set[role] != read_osd_set[role] && (start != 0 || end != chunk_size))
+            if (write_osd_set[role] != read_osd_set[role] && write_osd_set[role] != 0 && (start != 0 || end != chunk_size))
             {
                 // Copy new parity into the read buffer to write it back
                 memcpy(
-                    (uint8_t*)stripes[role].read_buf + start,
+                    (uint8_t*)stripes[role].read_buf + start - stripes[role].read_start,
                     stripes[role].write_buf,
                     end - start
                 );

src/osd_rmw_test.cpp

@@ -30,6 +30,7 @@ void test16();
 void test_recover_22_d2();
 void test_ec43_error_bruteforce();
 void test_recover_53_d5();
+void test_recover_22();
 
 int main(int narg, char *args[])
 {
@@ -70,6 +71,8 @@ int main(int narg, char *args[])
     test_ec43_error_bruteforce();
     // Test 19
     test_recover_53_d5();
+    // Test 20
+    test_recover_22();
     // End
     printf("all ok\n");
     return 0;
@@ -1244,3 +1247,99 @@ void test_recover_53_d5()
     // Done
     use_ec(8, 5, false);
 }
+
+void test_recover_22()
+{
+    const int bmp = 128*1024 / 4096 / 8;
+    use_ec(4, 2, true);
+    osd_num_t osd_set[4] = { 1, 2, 3, 4 };
+    osd_num_t write_osd_set[4] = { 5, 0, 3, 0 };
+    osd_rmw_stripe_t stripes[4] = {};
+    unsigned bitmaps[4] = { 0 };
+    // split
+    void *write_buf = (uint8_t*)malloc_or_die(4096);
+    set_pattern(write_buf, 4096, PATTERN0);
+    split_stripes(2, 128*1024, 120*1024, 4096, stripes);
+    assert(stripes[0].req_start == 120*1024 && stripes[0].req_end == 124*1024);
+    assert(stripes[1].req_start == 0 && stripes[1].req_end == 0);
+    assert(stripes[2].req_start == 0 && stripes[2].req_end == 0);
+    assert(stripes[3].req_start == 0 && stripes[3].req_end == 0);
+    // calc_rmw
+    void *rmw_buf = calc_rmw(write_buf, stripes, osd_set, 4, 2, 2, write_osd_set, 128*1024, bmp);
+    for (int i = 0; i < 4; i++)
+        stripes[i].bmp_buf = bitmaps+i;
+    assert(rmw_buf);
+    assert(stripes[0].read_start == 0 && stripes[0].read_end == 128*1024);
+    assert(stripes[1].read_start == 120*1024 && stripes[1].read_end == 124*1024);
+    assert(stripes[2].read_start == 0 && stripes[2].read_end == 0);
+    assert(stripes[3].read_start == 0 && stripes[3].read_end == 0);
+    assert(stripes[0].write_start == 120*1024 && stripes[0].write_end == 124*1024);
+    assert(stripes[1].write_start == 0 && stripes[1].write_end == 0);
+    assert(stripes[2].write_start == 120*1024 && stripes[2].write_end == 124*1024);
+    assert(stripes[3].write_start == 0 && stripes[3].write_end == 0);
+    assert(stripes[0].read_buf == (uint8_t*)rmw_buf+4*1024);
+    assert(stripes[1].read_buf == (uint8_t*)rmw_buf+132*1024);
+    assert(stripes[2].read_buf == NULL);
+    assert(stripes[3].read_buf == NULL);
+    assert(stripes[0].write_buf == write_buf);
+    assert(stripes[1].write_buf == NULL);
+    assert(stripes[2].write_buf == (uint8_t*)rmw_buf);
+    assert(stripes[3].write_buf == NULL);
+    // encode
+    set_pattern(stripes[0].read_buf, 128*1024, PATTERN1);
+    set_pattern(stripes[1].read_buf, 4*1024, PATTERN2);
+    memset(stripes[0].bmp_buf, 0xff, bmp);
+    memset(stripes[1].bmp_buf, 0xff, bmp);
+    calc_rmw_parity_ec(stripes, 4, 2, osd_set, write_osd_set, 128*1024, bmp);
+    assert(*(uint32_t*)stripes[2].bmp_buf == 0);
+    assert(stripes[0].write_start == 0 && stripes[0].write_end == 128*1024);
+    assert(stripes[1].write_start == 0 && stripes[1].write_end == 0);
+    assert(stripes[2].write_start == 120*1024 && stripes[2].write_end == 124*1024);
+    assert(stripes[3].write_start == 0 && stripes[3].write_end == 0);
+    assert(stripes[0].write_buf == stripes[0].read_buf);
+    assert(stripes[1].write_buf == NULL);
+    assert(stripes[2].write_buf == (uint8_t*)rmw_buf);
+    assert(stripes[3].write_buf == NULL);
+    check_pattern(stripes[2].write_buf, 4*1024, PATTERN0^PATTERN2);
+    // decode and verify
+    memset(stripes, 0, sizeof(stripes));
+    split_stripes(2, 128*1024, 0, 256*1024, stripes);
+    assert(stripes[0].req_start == 0 && stripes[0].req_end == 128*1024);
+    assert(stripes[1].req_start == 0 && stripes[1].req_end == 128*1024);
+    assert(stripes[2].req_start == 0 && stripes[2].req_end == 0);
+    assert(stripes[3].req_start == 0 && stripes[3].req_end == 0);
+    for (int role = 0; role < 4; role++)
+    {
+        stripes[role].read_start = stripes[role].req_start;
+        stripes[role].read_end = stripes[role].req_end;
+    }
+    assert(extend_missing_stripes(stripes, write_osd_set, 2, 4) == 0);
+    assert(stripes[0].read_start == 0 && stripes[0].read_end == 128*1024);
+    assert(stripes[1].read_start == 0 && stripes[1].read_end == 128*1024);
+    assert(stripes[2].read_start == 0 && stripes[2].read_end == 128*1024);
+    assert(stripes[3].read_start == 0 && stripes[3].read_end == 0);
+    void *read_buf = alloc_read_buffer(stripes, 4, 0);
+    for (int i = 0; i < 4; i++)
+        stripes[i].bmp_buf = bitmaps+i;
+    assert(read_buf);
+    assert(stripes[0].read_buf == read_buf);
+    assert(stripes[1].read_buf == (uint8_t*)read_buf+128*1024);
+    assert(stripes[2].read_buf == (uint8_t*)read_buf+2*128*1024);
+    set_pattern(stripes[0].read_buf, 128*1024, PATTERN1);
+    set_pattern(stripes[0].read_buf+120*1024, 4*1024, PATTERN0);
+    set_pattern(stripes[2].read_buf, 128*1024, PATTERN1^PATTERN2);
+    set_pattern(stripes[2].read_buf+120*1024, 4*1024, PATTERN0^PATTERN2);
+    memset(stripes[0].bmp_buf, 0xff, bmp);
+    memset(stripes[2].bmp_buf, 0, bmp);
+    bitmaps[1] = 0;
+    bitmaps[3] = 0;
+    reconstruct_stripes_ec(stripes, 4, 2, bmp);
+    assert(bitmaps[0] == 0xFFFFFFFF);
+    assert(*(uint32_t*)stripes[1].bmp_buf == 0xFFFFFFFF);
+    check_pattern(stripes[1].read_buf, 128*1024, PATTERN2);
+    free(read_buf);
+    // Done
+    free(rmw_buf);
+    free(write_buf);
+    use_ec(4, 2, false);
+}

src/osd_secondary.cpp

@@ -42,8 +42,10 @@ void osd_t::secondary_op_callback(osd_op_t *op)
     int retval = op->bs_op->retval;
     delete op->bs_op;
     op->bs_op = NULL;
-    if (op->is_recovery_related() && recovery_target_sleep_us)
+    if (op->is_recovery_related() && recovery_target_sleep_us &&
+        op->req.hdr.opcode == OSD_OP_SEC_STABILIZE)
     {
+        // Apply pause AFTER commit. Do not apply pause to SYNC at all
         if (!op->tv_end.tv_sec)
         {
             clock_gettime(CLOCK_REALTIME, &op->tv_end);
@@ -59,7 +61,25 @@ void osd_t::secondary_op_callback(osd_op_t *op)
     }
 }
 
-void osd_t::exec_secondary(osd_op_t *cur_op)
+void osd_t::exec_secondary(osd_op_t *op)
+{
+    if (op->is_recovery_related() && recovery_target_sleep_us &&
+        op->req.hdr.opcode != OSD_OP_SEC_STABILIZE && op->req.hdr.opcode != OSD_OP_SEC_SYNC)
+    {
+        // Apply pause BEFORE write/delete
+        tfd->set_timer_us(recovery_target_sleep_us, false, [this, op](int timer_id)
+        {
+            clock_gettime(CLOCK_REALTIME, &op->tv_begin);
+            exec_secondary_real(op);
+        });
+    }
+    else
+    {
+        exec_secondary_real(op);
+    }
+}
+
+void osd_t::exec_secondary_real(osd_op_t *cur_op)
 {
     if (cur_op->req.hdr.opcode == OSD_OP_SEC_READ_BMP)
     {

src/vitastor.pc.in

@@ -6,7 +6,7 @@ includedir=${prefix}/@CMAKE_INSTALL_INCLUDEDIR@
 
 Name: Vitastor
 Description: Vitastor client library
-Version: 1.3.1
+Version: 1.4.5
 Libs: -L${libdir} -lvitastor_client
 Cflags: -I${includedir}
 

tests/run_3osds.sh

@@ -10,6 +10,7 @@ SCHEME=${SCHEME:-replicated}
 # OFFSET_ARGS
 # PG_SIZE
 # PG_MINSIZE
+# GLOBAL_CONFIG
 
 if [ "$SCHEME" = "ec" ]; then
     OSD_COUNT=${OSD_COUNT:-5}
@@ -19,10 +20,10 @@ fi
 
 if [ "$IMMEDIATE_COMMIT" != "" ]; then
     NO_SAME="--journal_no_same_sector_overwrites true --journal_sector_buffer_count 1024 --disable_data_fsync 1 --immediate_commit all --log_level 10 --etcd_stats_interval 5"
-    $ETCDCTL put /vitastor/config/global '{"recovery_queue_depth":1,"recovery_tune_util_low":1,"osd_out_time":1,"immediate_commit":"all","client_enable_writeback":true}'
+    $ETCDCTL put /vitastor/config/global '{"recovery_queue_depth":1,"recovery_tune_util_low":1,"immediate_commit":"all","client_enable_writeback":true,"client_max_writeback_iodepth":32'$GLOBAL_CONFIG'}'
 else
     NO_SAME="--journal_sector_buffer_count 1024 --log_level 10 --etcd_stats_interval 5"
-    $ETCDCTL put /vitastor/config/global '{"recovery_queue_depth":1,"recovery_tune_util_low":1,"osd_out_time":1,"client_enable_writeback":true}'
+    $ETCDCTL put /vitastor/config/global '{"recovery_queue_depth":1,"recovery_tune_util_low":1,"client_enable_writeback":true,"client_max_writeback_iodepth":32'$GLOBAL_CONFIG'}'
 fi
 
 start_osd_on()
@@ -53,7 +54,7 @@ for i in $(seq 1 $OSD_COUNT); do
     start_osd $i
 done
 
-(while true; do node mon/mon-main.js --etcd_address $ETCD_URL --etcd_prefix "/vitastor" --verbose 1 || true; done) >>./testdata/mon.log 2>&1 &
+(while true; do set +e; node mon/mon-main.js --etcd_address $ETCD_URL --etcd_prefix "/vitastor" --verbose 1; if [[ $? -ne 2 ]]; then break; fi; done) >>./testdata/mon.log 2>&1 &
 MON_PID=$!
 
 if [ "$SCHEME" = "ec" ]; then

tests/run_tests.sh

@@ -45,6 +45,8 @@ IMMEDIATE_COMMIT=1 ./test_rebalance_verify.sh
 SCHEME=ec ./test_rebalance_verify.sh
 SCHEME=ec IMMEDIATE_COMMIT=1 ./test_rebalance_verify.sh
 
+./test_switch_primary.sh
+
 ./test_write.sh
 SCHEME=xor ./test_write.sh
 

tests/test_add_osd.sh

@@ -1,7 +1,7 @@
 #!/bin/bash -ex
 
 PG_COUNT=2048
-
+GLOBAL_CONFIG=',"osd_out_time":1'
 . `dirname $0`/run_3osds.sh
 
 LD_PRELOAD="build/src/libfio_vitastor.so" \

tests/test_heal.sh

@@ -9,6 +9,7 @@ if [[ "$SCHEME" = "ec" ]]; then
 fi
 OSD_COUNT=${OSD_COUNT:-7}
 PG_COUNT=32
+GLOBAL_CONFIG=',"osd_out_time":1'
 . `dirname $0`/run_3osds.sh
 check_qemu
 

tests/test_minsize_1.sh

@@ -2,6 +2,7 @@
 
 PG_MINSIZE=1
 SCHEME=replicated
+GLOBAL_CONFIG=',"osd_out_time":1'
 
 . `dirname $0`/run_3osds.sh
 

tests/test_scrub.sh

@@ -20,6 +20,9 @@ LD_PRELOAD="build/src/libfio_vitastor.so" \
     fio -thread -name=test -ioengine=build/src/libfio_vitastor.so -bs=1M -direct=1 -iodepth=4 \
         -mirror_file=./testdata/mirror.bin -end_fsync=1 -rw=write -etcd=$ETCD_URL -image=testimg
 
+# Save PG primary
+primary=$($ETCDCTL get --print-value-only /vitastor/config/pgs | jq -r '.items["1"]["1"].primary')
+
 # Intentionally corrupt OSD data and restart it
 zero_osd_pid=OSD${ZERO_OSD}_PID
 kill ${!zero_osd_pid}
@@ -34,6 +37,9 @@ start_osd $ZERO_OSD
 # Wait until start
 wait_up 10
 
+# Wait until PG is back on the same primary
+wait_condition 10 "$ETCDCTL"$' get --print-value-only /vitastor/config/pgs | jq -s -e \'.[0].items["1"]["1"].primary == "'$primary'"'"'"
+
 # Trigger scrub
 $ETCDCTL put /vitastor/pg/history/1/1 `$ETCDCTL get --print-value-only /vitastor/pg/history/1/1 | jq -s -c '(.[0] // {}) + {"next_scrub":1}'`
 

tests/test_splitbrain.sh

@@ -4,6 +4,7 @@ OSD_COUNT=2
 PG_SIZE=2
 PG_MINSIZE=1
 SCHEME=replicated
+GLOBAL_CONFIG=',"osd_out_time":1'
 
 . `dirname $0`/run_3osds.sh
 

tests/test_switch_primary.sh

@@ -0,0 +1,18 @@
+#!/bin/bash -ex
+
+. `dirname $0`/run_3osds.sh
+
+primary=$($ETCDCTL get --print-value-only /vitastor/config/pgs | jq -r '.items["1"]["1"].primary')
+primary_pid=OSD${primary}_PID
+kill -9 ${!primary_pid}
+
+sleep 15
+wait_condition 10 "$ETCDCTL get --print-value-only /vitastor/config/pgs | jq -s -e '.[0].items[\"1\"][\"1\"].primary != \"$primary\"'"
+
+newprim=$($ETCDCTL get --print-value-only /vitastor/config/pgs | jq -r '.items["1"]["1"].primary')
+
+if [ "$newprim" = "$primary" ]; then
+    format_error Primary not switched
+fi
+
+format_green OK