[
https://issues.apache.org/jira/browse/IGNITE-21220?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17814163#comment-17814163
]
Dmitry Baranov commented on IGNITE-21220:
-----------------------------------------
In case of network partitioning node will be removed from logical topology, to
join topology again node need to be restarted, thus jobs will be canceled
during restart. There is no need to process jobs on initial worker node in case
of onNodeLeft event
> Worker node recovery
> --------------------
>
> Key: IGNITE-21220
> URL: https://issues.apache.org/jira/browse/IGNITE-21220
> Project: Ignite
> Issue Type: Improvement
> Reporter: Aleksandr
> Priority: Major
> Labels: ignite-3
>
> There is a case when the worker node executing some job has left the logical
> topology. This node has to identify this situation and stop all running jobs.
> We must not have the situation when the worker left the topology, the
> coordinator restarted the job on another node, then the first worker joined
> the topology and two instances of the job are running on the cluster.
> Maybe we do not allow this but we have a lack of tests here.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
