[ 
https://issues.apache.org/jira/browse/IGNITE-24363?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Roman Puchkovskiy reassigned IGNITE-24363:
------------------------------------------

    Assignee: Roman Puchkovskiy

> Implement TxCleanupRecoveryRequest processing for zone replica
> --------------------------------------------------------------
>
>                 Key: IGNITE-24363
>                 URL: https://issues.apache.org/jira/browse/IGNITE-24363
>             Project: Ignite
>          Issue Type: Improvement
>            Reporter: Alexander Lapin
>            Assignee: Roman Puchkovskiy
>            Priority: Major
>              Labels: ignite-3
>
> h3. Motivation
> Within colocation track we should reasonably move table-not-specific requests 
> and commands processing from PartitionReplicaListener and PartitionListener 
> to ZonePartitionReplicaListener and ZonePartitionRaftListener. 
> TxCleanupRecoveryRequest is one of such requests to be moved. In this 
> particular case TxRecoveryMessage is expected to be moved.
> h3. Definition of Done
>  * Transaction recovery triggered by primary replica election works properly.
> h3. Implementation Notes
>  * Initial source for groupId to recover is PRIMARY_REPLICA_ELECTED See 
> TxManagerImpl#primaryReplicaElectedListener for more details. Thus, besides 
> ensuring that corresponding event is fired in case of zone primary replica 
> (meaning checking that TestPlacementDriver does fire such event) it's also 
> required to properly convert TablePartitionId to ZonePartitionId like we did 
> previously in similar tickets.
> {code:java}
> .groupId(toTablePartitionIdMessage(REPLICA_MESSAGES_FACTORY, 
> tablePartitionId)){code}
>  * Copy PartitionReplicaListener#processCleanupRecoveryMessage along with 
> corresponding trigger to  ZonePartitionReplicaListener.
>  * Hide original PartitionReplicaListener#processCleanupRecoveryMessage under 
> colocation feature flag.
>  * Implement corresponding smoke test to ensure that transaction recovery 
> works properly.



--
This message was sent by Atlassian Jira
(v8.20.10#820010)

Reply via email to