You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@kudu.apache.org by "Rodion Myronov (JIRA)" <ji...@apache.org> on 2018/05/24 10:35:00 UTC
[jira] [Commented] (KUDU-1779) Consensus "stuck" with all
transaction trackers are at limit
[ https://issues.apache.org/jira/browse/KUDU-1779?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16488776#comment-16488776 ]
Rodion Myronov commented on KUDU-1779:
--------------------------------------
Faced this issue in my cluster. Resolved by cluster restart.
[~tlipcon], may I ask for your opinion regarding using --tablet_transaction_memory_limit_mb as a workaround for this issue? Would it be useful to increase it (from default 64MB) or set to -1 to disable tracking at all? What negative consequences should I expect after such change?
> Consensus "stuck" with all transaction trackers are at limit
> ------------------------------------------------------------
>
> Key: KUDU-1779
> URL: https://issues.apache.org/jira/browse/KUDU-1779
> Project: Kudu
> Issue Type: Bug
> Components: consensus
> Affects Versions: 1.1.0
> Reporter: Todd Lipcon
> Assignee: Todd Lipcon
> Priority: Critical
> Labels: stability
>
> In a stress cluster, I saw one tablet get "stuck" in the following state:
> - the transaction_tracker on all three replicas is "full" (no more can be submitted)
> - leader elections proceed just fine, but no leader is able to advance the commit index
> The issue seems to be that a replica will respond with 'CANNOT_PREPARE' when its transaction tracker is full. The leader then ignores this response, and doesn't advance the majority-replicated watermark. The transaction tracker stays full forever because the in-flight transactions can't get committed.
> Notes to follow.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)