[
https://issues.apache.org/jira/browse/IGNITE-25126?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17944645#comment-17944645
]
Roman Puchkovskiy commented on IGNITE-25126:
--------------------------------------------
After we initiate peers reset, we wait for the leader to be established.
According to the logs, this step was probably the one where something 'stuck'
for a long time. This step is not needed actually, so it seems that we can just
remove it.
> repairWhenMgWas3Nodes is flaky
> ------------------------------
>
> Key: IGNITE-25126
> URL: https://issues.apache.org/jira/browse/IGNITE-25126
> Project: Ignite
> Issue Type: Improvement
> Reporter: Roman Puchkovskiy
> Assignee: Roman Puchkovskiy
> Priority: Major
> Labels: ignite-3
> Attachments:
> _Integration_Tests_Module_System_Disaster_Recovery_32.log.zip
>
> Time Spent: 10m
> Remaining Estimate: 0h
>
> In
> [https://ci.ignite.apache.org/buildConfiguration/ApacheIgnite3xGradle_Test_IntegrationTests_ModuleSystemDisasterRecovery/9039040]
> , it can be seen in the logs that MG repair process was initiated, that
> lonely leader was appointed and that it successfully extended the
> configuration to the desired voting set consisting of 3 nodes. But the
> becomeLonelyLeader()'s future did not complete in 10 seconds for some reason.
> The suspect is the process of getting the leader after initiating the
> forceful configuration change. We need more logs to get to the bottom of this.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
