You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@ignite.apache.org by "Mikhail Efremov (Jira)" <ji...@apache.org> on 2024/12/10 10:35:00 UTC
[jira] [Updated] (IGNITE-22372) Moving RaftManager into Replica
[ https://issues.apache.org/jira/browse/IGNITE-22372?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Mikhail Efremov updated IGNITE-22372:
-------------------------------------
Epic Link: IGNITE-23862 (was: IGNITE-22313)
> Moving RaftManager into Replica
> -------------------------------
>
> Key: IGNITE-22372
> URL: https://issues.apache.org/jira/browse/IGNITE-22372
> Project: Ignite
> Issue Type: Improvement
> Reporter: Mikhail Efremov
> Assignee: Mikhail Efremov
> Priority: Major
> Labels: ignite-3
>
> *Description*
> The main goal is to move {{RaftManager}} into {{Replica}} through it's constructor and starting raft-node and raft-client in {{Replica}} and stopping on {{Replica}}'s shutdown process inside a corresponding replica.
> *Motivation*
> After IGNITE-21805 {{RaftManager}} was moved from {{TableManager}} into {{ReplicaManager}}. Therefore raft-nodes and raft-replicas are starting and stopping inside {{ReplicaManager}}. The main idea of previous ticket is to hide all RAFT-related entities as far as possible into Replica itself. Then, {{RaftManager}} and related usages should be moved into {{ReplicaManager.}}
> *Definition of Done*
> # {{RaftManager}} component is moved to {{Replica}}.
> # {{Loza#startRaftGroupNode}} and {{RaftManager#stopRaftNodes}} should be called inside {{Replica}}.
> *Current issues*
> There no of determination how to handle this code: {{((Loza) raftManager).volatileRaft().logStorage().value();}} In theory, the only purpose is to create raft options and pass them into {{Loza#startRaftGroupNode}} and critical challenges shouldn't arise there. But the attention is here.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)