[
https://issues.apache.org/jira/browse/IGNITE-27185?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Filipp Shergalis updated IGNITE-27185:
--------------------------------------
Description:
Streaming snapshot without any commands applied to log after it seems to be
broken in RAFT, so we try to send it repeatedly, getting "
Register DownloadingSnapshot failed: snapshot is not newer"
Not relevant to CMG tests, looks like CMG group applies some commands after
snapshot is taken and before shutting down the cluster, so the problem is not
reproduced
Updating TruncateRaftLogCommand to make 1 snapshot and just using this job
twice to cause log truncation seems to fix the issue with the test (not sure
why)
was:
Streaming snapshot without any commands applied to log after it seems to be
broken in RAFT, so we try to send it repeatedly, getting "
Register DownloadingSnapshot failed: snapshot is not newer"
Not relevant to CMG tests, looks like CMG group applies some commands after
snapshot is taken and before shutting down the cluster, so the problem is not
reproduced
Updating TruncateRaftLogCommand to make 1 snapshot and just using this job
twice to cause log truncation seems to fix the issue with the test (not sure
why)
> ItMetastorageRaftSnapshotCompatibilityTest is broken
> ----------------------------------------------------
>
> Key: IGNITE-27185
> URL: https://issues.apache.org/jira/browse/IGNITE-27185
> Project: Ignite
> Issue Type: Improvement
> Reporter: Roman Puchkovskiy
> Priority: Major
> Labels: MakeTeamcityGreenAgain, ignite-3
> Fix For: 3.2
>
>
> Streaming snapshot without any commands applied to log after it seems to be
> broken in RAFT, so we try to send it repeatedly, getting "
> Register DownloadingSnapshot failed: snapshot is not newer"
>
> Not relevant to CMG tests, looks like CMG group applies some commands after
> snapshot is taken and before shutting down the cluster, so the problem is not
> reproduced
>
> Updating TruncateRaftLogCommand to make 1 snapshot and just using this job
> twice to cause log truncation seems to fix the issue with the test (not sure
> why)
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
