[
https://issues.apache.org/jira/browse/IGNITE-24193?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Alexander Lapin updated IGNITE-24193:
-------------------------------------
Epic Link: IGNITE-24953 (was: IGNITE-22115)
> Enchance PartitionReplicaLifecycleManager and ZonePartitionReplicaImpl with
> primary-based replication mechanism
> ---------------------------------------------------------------------------------------------------------------
>
> Key: IGNITE-24193
> URL: https://issues.apache.org/jira/browse/IGNITE-24193
> Project: Ignite
> Issue Type: Improvement
> Reporter: Mikhail Efremov
> Priority: Major
> Labels: ignite-3
>
> *Description*
> After IGNITE-22313 we moved replication actor's business from a partition
> RAFT-group leader to the partition primary replica. In particular now there a
> new {{ChangePeersAndLearnersAsyncReplicaRequest}} that is sent from
> {{TableManager}}. This logic wasn't exported to colocation's entities e.g.
> {{PartitionReplicaLifecycleManager}}.
> *Motivation*
> We shouldn't loose new rebalance logic during zone colocation epic
> IGNITE-22115.
> *Definition of done*
> # Zone replicas are able to handle primary replica elected events with
> {{changePeersAndLearnersAsync}} calls alongside raft group leader elected
> callback subscription.
> # {{PartitionReplicaLifecycleManager}} sends
> {{ChangePeersAndLearnersAsyncReplicaRequest}} after zone pending assignments
> handling done as it was done in {{TableManager}}.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
