You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@ignite.apache.org by "Sergey Uttsel (Jira)" <ji...@apache.org> on 2021/12/16 08:54:00 UTC
[jira] [Comment Edited] (IGNITE-16097) Some tests are flaky in ItNodeTest
[ https://issues.apache.org/jira/browse/IGNITE-16097?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17458168#comment-17458168 ]
Sergey Uttsel edited comment on IGNITE-16097 at 12/16/21, 8:53 AM:
-------------------------------------------------------------------
[https://github.com/apache/ignite-3/pull/501]
All these tests fail due to excessive re-election of the leader.
# Most of the tests were fixed by increasing the election timeout to 1000ms.
# testTripleNodesWithReplicatorStateListener and testLeaderTransferWithReplicatorStateListener fail due to a bug introduced in the ticket https://issues.apache.org/jira/browse/IGNITE-15528.
The current term must be read under the writeLock. But after the changes in IGNITE-15528, it began to be read in asynchronous code, without writeLock.
[~ascherbakov] please review it.
was (Author: sergey uttsel):
[https://github.com/apache/ignite-3/pull/501]
Increased election timeout. In some tests, I used stoppedCounter, since sometimes the leader changes several times at the start of a nodes.
> Some tests are flaky in ItNodeTest
> ----------------------------------
>
> Key: IGNITE-16097
> URL: https://issues.apache.org/jira/browse/IGNITE-16097
> Project: Ignite
> Issue Type: Task
> Reporter: Sergey Uttsel
> Assignee: Sergey Uttsel
> Priority: Major
> Labels: ignite-3
>
> Some tests are flaky in org.apache.ignite.raft.jraft.core.ItNodeTest:
> # testLeaderTransferWithReplicatorStateListener
> # testTripleNodesWithReplicatorStateListener
> # testResetLearners
> # testLeaderTransfer
> # testLeaderTransferBeforeLogIsCompleted
> # testLeaderTransferResumeOnFailure
>
--
This message was sent by Atlassian Jira
(v8.20.1#820001)