[jira] [Commented] (IGNITE-17911) Wal isn't enabled for some caches after cancelling of rebalance

["Ignite TC Bot (Jira)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/IGNITE-17911?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17619047#comment-17619047 ] 

Ignite TC Bot commented on IGNITE-17911:
----------------------------------------

{panel:title=Branch: [pull/10320/head] Base: [master] : No blockers found!|borderStyle=dashed|borderColor=#ccc|titleBGColor=#D6F7C1}{panel}
{panel:title=Branch: [pull/10320/head] Base: [master] : New Tests (1)|borderStyle=dashed|borderColor=#ccc|titleBGColor=#D6F7C1}
{color:#00008b}PDS 4{color} [[tests 1|https://ci.ignite.apache.org/viewLog.html?buildId=6839956]]
* {color:#013220}IgnitePdsTestSuite4: IgniteDisableWalOnRebalanceTest.testDisabledWalOnRebalance - PASSED{color}

{panel}
[TeamCity *--> Run :: All* Results|https://ci.ignite.apache.org/viewLog.html?buildId=6839222&buildTypeId=IgniteTests24Java8_RunAll]

> Wal isn't enabled for some caches after cancelling of rebalance
> ---------------------------------------------------------------
>
>                 Key: IGNITE-17911
>                 URL: https://issues.apache.org/jira/browse/IGNITE-17911
>             Project: Ignite
>          Issue Type: Task
>            Reporter: Aleksandr Polovtcev
>            Assignee: Aleksandr Polovtcev
>            Priority: Major
>          Time Spent: 10m
>  Remaining Estimate: 0h
>
> WAL can be disabled during a rebalance if a node does not own any partitions. When stopping a node, a shutdown hook is used, which calls {{IgniteionEx#stop}} with the {{cancel}} flag set to {{true}}. This wakes up the checkpoint thread and starts doing a checkpoint, which creates a checkpoint start marker. However, since the {{cancel}} flag was set to {{true}}, {{Checkpointer#writePages}} finishes immediately and the checkpoint end marker is not created.
> This means that we have not enabled WAL again, since the rebalance was interrupted, and we created a checkpoint start marker, but not the end marker. This leads to the node being started in maintenance mode.



--
This message was sent by Atlassian Jira
(v8.20.10#820010)