[
https://issues.apache.org/jira/browse/IGNITE-26763?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Mirza Aliev updated IGNITE-26763:
---------------------------------
Description:
h3. Motivation
Currently {{PartitionReplicaLifecycleManager#stopPartitionInternal}} is
implemented in the way that such operation supposed to be performed only in
node stop flow:
{code:java}
// Not using the busy lock here, because this method is called on component
stop.
{code}
Now this assumption is wrong, because partition restart/restart with cleanup
reuses such logic and it could be performed not only in node stop flow.
We need to cover {{#stopPartition}} with corresponding busy locks.
was:
h3. Motivation
Currently {{PartitionReplicaLifecycleManager#stopPartitionInternal}} is
implemented in the way that such operation supposed to be performed only in
node stop flow:
{code:java}
// Not using the busy lock here, because this method is called on component
stop.
{code}
Now this assumption is wrong, because partition restart/restart with cleanup
reuses such logic and it could be performed not only in node stop flow.
We need to cover {{#stopPartition}} with corresponding busy locks.
> Improve busy lock coverage for the stop partition flow
> ------------------------------------------------------
>
> Key: IGNITE-26763
> URL: https://issues.apache.org/jira/browse/IGNITE-26763
> Project: Ignite
> Issue Type: Improvement
> Reporter: Mirza Aliev
> Assignee: Mirza Aliev
> Priority: Major
> Labels: ignite-3
>
> h3. Motivation
> Currently {{PartitionReplicaLifecycleManager#stopPartitionInternal}} is
> implemented in the way that such operation supposed to be performed only in
> node stop flow:
> {code:java}
> // Not using the busy lock here, because this method is called on component
> stop.
> {code}
>
> Now this assumption is wrong, because partition restart/restart with cleanup
> reuses such logic and it could be performed not only in node stop flow.
> We need to cover {{#stopPartition}} with corresponding busy locks.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
