You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@ignite.apache.org by "Pavel Tupitsyn (Jira)" <ji...@apache.org> on 2026/03/19 12:57:00 UTC
[jira] [Commented] (IGNITE-27651) Implement client liveness detection for a directly mapped transaction
[ https://issues.apache.org/jira/browse/IGNITE-27651?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18066784#comment-18066784 ]
Pavel Tupitsyn commented on IGNITE-27651:
-----------------------------------------
https://github.com/apache/ignite-3/commit/3eb50900656064276d76fc346d4e90ce7765e892
> Implement client liveness detection for a directly mapped transaction
> ---------------------------------------------------------------------
>
> Key: IGNITE-27651
> URL: https://issues.apache.org/jira/browse/IGNITE-27651
> Project: Ignite
> Issue Type: Improvement
> Components: platforms ai3, transactions ai3
> Reporter: Alexey Scherbakov
> Assignee: Pavel Tupitsyn
> Priority: Major
> Labels: ignite-3
> Fix For: 3.2
>
> Time Spent: 1h 10m
> Remaining Estimate: 0h
>
> Currently directly mapped transactions rely on transaction timeouts to cleanup locks in case of client failure (which serves as a lightweight coordinator).
> This will not work in case of very big timeouts.
> We need to implement some kind of liveness detection for a client and initiate tx state resolution in case of client's death.
> A quick solution would be limiting the upper bound for tx timeouts.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)