You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@ratis.apache.org by "Tsz Wo Nicholas Sze (JIRA)" <ji...@apache.org> on 2018/12/01 01:48:00 UTC
[jira] [Commented] (RATIS-443) FollowerState.inLogSync can be
incorrectly cleared in appendEntriesAsync
[ https://issues.apache.org/jira/browse/RATIS-443?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16705569#comment-16705569 ]
Tsz Wo Nicholas Sze commented on RATIS-443:
-------------------------------------------
r443_20181130.patch: counts the number of outstanding ops.
> FollowerState.inLogSync can be incorrectly cleared in appendEntriesAsync
> ------------------------------------------------------------------------
>
> Key: RATIS-443
> URL: https://issues.apache.org/jira/browse/RATIS-443
> Project: Ratis
> Issue Type: Bug
> Components: server
> Reporter: Tsz Wo Nicholas Sze
> Assignee: Tsz Wo Nicholas Sze
> Priority: Major
> Attachments: r443_20181130.patch
>
>
> FollowerState.inLogSync is to indicate that the follower is writing log/statemachine data. So that it should not time out to start a leader election. However, it is not working correctly in the async case: suppose there are two or more outstanding appendEntriesAsync calls, the first completed call will clear inLogSync even if the other calls are still in progress.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)