[ 
https://issues.apache.org/jira/browse/IGNITE-24558?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17928519#comment-17928519
 ] 

Vyacheslav Koptilin commented on IGNITE-24558:
----------------------------------------------

The root cause of the issue is a race between triggering 
`UpdateMinimumActiveTxBeginTimeReplicaRequest` and checkpoint.
When the request is triggered before the checkpoint, updating ` the 
MinimumRequiredTimeCollectorService` is postponed until the next checkpoint, 
which is 3 minutes by default. In that case, catalog compaction cannot truncate 
the catalog's history.

> ItReplicaLifecycleTest.testCatalogCompaction is flaky
> -----------------------------------------------------
>
>                 Key: IGNITE-24558
>                 URL: https://issues.apache.org/jira/browse/IGNITE-24558
>             Project: Ignite
>          Issue Type: Bug
>            Reporter: Alexander Lapin
>            Assignee: Vyacheslav Koptilin
>            Priority: Major
>              Labels: ignite-3
>          Time Spent: 10m
>  Remaining Estimate: 0h
>
> {code:java}
> org.opentest4j.AssertionFailedError: Failed to wait for the expected catalog 
> version [expected=3, earliest=2, latest=4] ==> expected: <true> but was: 
> <false>  at 
> app//org.junit.jupiter.api.AssertionFailureBuilder.build(AssertionFailureBuilder.java:151)
>   at 
> app//org.junit.jupiter.api.AssertionFailureBuilder.buildAndThrow(AssertionFailureBuilder.java:132)
>   at app//org.junit.jupiter.api.AssertTrue.failNotTrue(AssertTrue.java:63)  
> at app//org.junit.jupiter.api.AssertTrue.assertTrue(AssertTrue.java:36)  at 
> app//org.junit.jupiter.api.Assertions.assertTrue(Assertions.java:214)  at 
> app//org.apache.ignite.internal.partition.replicator.ItReplicaLifecycleTest.expectEarliestCatalogVersion(ItReplicaLifecycleTest.java:716)
>   at 
> app//org.apache.ignite.internal.partition.replicator.ItReplicaLifecycleTest.testCatalogCompaction(ItReplicaLifecycleTest.java:710)
>  {code}
> Reproduced locally 3/10 both on main and 
> [https://github.com/apache/ignite-3/pull/5230.] Seems that flaky rate on 5230 
> is higher.



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

Reply via email to