[
https://issues.apache.org/jira/browse/IGNITE-24849?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Aleksandr Polovtsev updated IGNITE-24849:
-----------------------------------------
Description: After a remote Raft snapshot has been installed on a node, it
must flush its storages in order to make the installed snapshot durable.
Otherwise, on node restart, Raft node may fail to replay its log in case the
installed snapshot was not flushed, because we will tell JRaft to restore data
from the 0 index, while not having the full log locally (that's why a snapshot
was installed on this node in the first place). (was: After a remote Raft
snapshot has been installed on a node, it must flush its storages in order to
make the installed snapshot durable. Otherwise, on node restart, Raft node may
fail to replay its log in case the installed snapshot was not flushed, because
the log prefix will also be absent in that case.)
> Flush storages after rebalance
> ------------------------------
>
> Key: IGNITE-24849
> URL: https://issues.apache.org/jira/browse/IGNITE-24849
> Project: Ignite
> Issue Type: Bug
> Reporter: Aleksandr Polovtsev
> Assignee: Aleksandr Polovtsev
> Priority: Major
> Labels: ignite-3
>
> After a remote Raft snapshot has been installed on a node, it must flush its
> storages in order to make the installed snapshot durable. Otherwise, on node
> restart, Raft node may fail to replay its log in case the installed snapshot
> was not flushed, because we will tell JRaft to restore data from the 0 index,
> while not having the full log locally (that's why a snapshot was installed on
> this node in the first place).
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
