You are viewing a plain text version of this content. The canonical link for it is here.

Posted to issues@mesos.apache.org by "Jake Farrell (JIRA)" <ji...@apache.org> on 2015/05/13 05:25:00 UTC

[jira] [Updated] (MESOS-419) Old slave directories should not be garbage collected based on file modification time

     [ https://issues.apache.org/jira/browse/MESOS-419?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Jake Farrell updated MESOS-419:
-------------------------------
    Issue Type: Bug  (was: Sub-task)
        Parent:     (was: MESOS-308)

> Old slave directories should not be garbage collected based on file modification time
> -------------------------------------------------------------------------------------
>
>                 Key: MESOS-419
>                 URL: https://issues.apache.org/jira/browse/MESOS-419
>             Project: Mesos
>          Issue Type: Bug
>            Reporter: Vinod Kone
>            Assignee: Vinod Kone
>             Fix For: 0.12.0
>
>
> Since running executors, do not change the modification time of the slave work directory, it would be bad to schedule old slave directories for deletion based on their mod time. Instead, these work directories should be scheduled for gc based on the 'gc_delay' flag.
> At Twitter, this manifested as task directories getting immediately gc'ed when long running slaves were rolled for an upgrade.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)