You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@mesos.apache.org by "Yan Xu (JIRA)" <ji...@apache.org> on 2014/01/04 00:03:51 UTC
[jira] [Commented] (MESOS-892) Additional Issues with contender
related change
[ https://issues.apache.org/jira/browse/MESOS-892?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13861997#comment-13861997 ]
Yan Xu commented on MESOS-892:
------------------------------
1. https://reviews.apache.org/r/16586/
2. https://reviews.apache.org/r/16588/
3. https://reviews.apache.org/r/16589/
4. https://reviews.apache.org/r/16590/
5. https://reviews.apache.org/r/16591/
> Additional Issues with contender related change
> -----------------------------------------------
>
> Key: MESOS-892
> URL: https://issues.apache.org/jira/browse/MESOS-892
> Project: Mesos
> Issue Type: Bug
> Reporter: Yan Xu
> Assignee: Yan Xu
> Fix For: 0.16.0
>
>
> There are a few issues with the contender code in master, mostly with regard to semantic consistency of the APIs.
> 1. When the detectors (ZooKeeperMasterDetector, LeaderDetector) fail, subsequent detect() calls should get a Failure directly.
> 2. ZooKeeperMasterContenderProcess::master should be an option to codify the fact that it can be unset after construction.
> 3. ZooKeeperMasterContenderProcess::contend should check if the future for the previous contend() is still pending and if so it should directly return that future without recontending.
> 4. LeaderContender::withdraw() should alway return ‘false’ if there is no membership to withdraw.
> 5. GroupProcess::abort() should set all owned memberships as cancelled.
--
This message was sent by Atlassian JIRA
(v6.1.5#6160)