[
https://issues.apache.org/jira/browse/IGNITE-12548?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17019408#comment-17019408
]
Ivan Rakov commented on IGNITE-12548:
-------------------------------------
[~ascherbakov] Looks good.
> Possible tx desync during recovery on near node left.
> -----------------------------------------------------
>
> Key: IGNITE-12548
> URL: https://issues.apache.org/jira/browse/IGNITE-12548
> Project: Ignite
> Issue Type: Bug
> Affects Versions: 2.7.6
> Reporter: Alexey Scherbakov
> Assignee: Alexey Scherbakov
> Priority: Blocker
> Fix For: 2.9
>
> Time Spent: 10m
> Remaining Estimate: 0h
>
> The problem appears if a transaction is starting to rollback in PREPARED
> state for some reason and concurrently near node is left triggering tx
> recovery protocol.
> Consider having two enlisted keys from different partitions mapped to
> different nodes N1 and N2.
> Due to race N1 local tx can be rolled back while N2 local tx is committed
> breaking tx atomicity guarantee.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
