[jira] [Updated] (IGNITE-24491) Incorrect zone replica stop order

["Aleksandr Polovtsev (Jira)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/IGNITE-24491?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Aleksandr Polovtsev updated IGNITE-24491:
-----------------------------------------
    Description: 
In the colocation track, on node stop, zone-wide replica stop happens in the following steps:
# {{ReplicaManager}} calls {{Replica#shutdown}};
# {{ZonePartitionReplicaImpl#shutdown}} closes the zone-partition TX storage;
# {{ReplicaManager}} stops the Raft node.

This is a problem, because Raft may also need to access the TX storage. The correct order is "stop replica -> stop raft -> stop TX storage". 

> Incorrect zone replica stop order
> ---------------------------------
>
>                 Key: IGNITE-24491
>                 URL: https://issues.apache.org/jira/browse/IGNITE-24491
>             Project: Ignite
>          Issue Type: Bug
>            Reporter: Aleksandr Polovtsev
>            Assignee: Aleksandr Polovtsev
>            Priority: Major
>              Labels: ignite-3
>
> In the colocation track, on node stop, zone-wide replica stop happens in the following steps:
> # {{ReplicaManager}} calls {{Replica#shutdown}};
> # {{ZonePartitionReplicaImpl#shutdown}} closes the zone-partition TX storage;
> # {{ReplicaManager}} stops the Raft node.
> This is a problem, because Raft may also need to access the TX storage. The correct order is "stop replica -> stop raft -> stop TX storage". 



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