You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@slider.apache.org by "Steve Loughran (JIRA)" <ji...@apache.org> on 2014/08/11 12:40:12 UTC
[jira] [Commented] (SLIDER-78) add a programmable delay before
re-issuing a request for a container after a failure
[ https://issues.apache.org/jira/browse/SLIDER-78?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14092659#comment-14092659 ]
Steve Loughran commented on SLIDER-78:
--------------------------------------
now that we have a scheduled task queue, this is straightforward to add.
But do we need to, now that the agents are making more decisions on the target machines, including port-scanning?
> add a programmable delay before re-issuing a request for a container after a failure
> ------------------------------------------------------------------------------------
>
> Key: SLIDER-78
> URL: https://issues.apache.org/jira/browse/SLIDER-78
> Project: Slider
> Issue Type: Sub-task
> Components: appmaster
> Reporter: Steve Loughran
> Assignee: Sumit Mohanty
> Priority: Minor
>
> A failure mode that surfaces is restarting a cluster immediately after freezing it. The port the HBase master wanted had not yet been released by the OS, so it failed when trying to listen on it -repeatedly.
> Having a delay on execing the actual binary (i.e. get the container, but insert a sleep) would address this
--
This message was sent by Atlassian JIRA
(v6.2#6252)