[
https://issues.apache.org/jira/browse/IGNITE-19833?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Vladislav Pyatkov updated IGNITE-19833:
---------------------------------------
Description:
*Motivation*
In cases where repeating the RAFT command does not guarantee a result, we want
to know that and have the possibility of waiting for the recovery majority.
Instead of continuing the sanding request and receiving a timeout.
*Example*
A replica is handling a lease grant message. In that case, the replica should
request a leader index from RAFT. Of course, if the majority is already lost,
the index result won't be received immediately. As a result, the RAFT client
must notify about it in some way and also provide a way to wait for the
majority. In this case, the waiting would be tied to the lease grant message
expiration time.
> Function to understand that the replication group majority is lost
> ------------------------------------------------------------------
>
> Key: IGNITE-19833
> URL: https://issues.apache.org/jira/browse/IGNITE-19833
> Project: Ignite
> Issue Type: Improvement
> Reporter: Vladislav Pyatkov
> Priority: Major
> Labels: ignite-3
>
> *Motivation*
> In cases where repeating the RAFT command does not guarantee a result, we
> want to know that and have the possibility of waiting for the recovery
> majority. Instead of continuing the sanding request and receiving a timeout.
> *Example*
> A replica is handling a lease grant message. In that case, the replica should
> request a leader index from RAFT. Of course, if the majority is already lost,
> the index result won't be received immediately. As a result, the RAFT client
> must notify about it in some way and also provide a way to wait for the
> majority. In this case, the waiting would be tied to the lease grant message
> expiration time.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
