You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@mesos.apache.org by "Benjamin Bannier (JIRA)" <ji...@apache.org> on 2018/08/15 13:48:00 UTC
[jira] [Commented] (MESOS-9156) StorageLocalResourceProviderProcess
can deadlock
[ https://issues.apache.org/jira/browse/MESOS-9156?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16581086#comment-16581086 ]
Benjamin Bannier commented on MESOS-9156:
-----------------------------------------
Review: https://reviews.apache.org/r/68361/
> StorageLocalResourceProviderProcess can deadlock
> ------------------------------------------------
>
> Key: MESOS-9156
> URL: https://issues.apache.org/jira/browse/MESOS-9156
> Project: Mesos
> Issue Type: Bug
> Components: agent
> Affects Versions: 1.7.0
> Reporter: Benjamin Bannier
> Assignee: Benjamin Bannier
> Priority: Major
> Attachments: log
>
>
> On fatal conditions the {{StorageLocalResourceProviderProcess}} triggers its {{fatal}} function which causes its {{Driver}} process to be torn down. Invocations of {{fatal}} need to be properly {{defer}}'d and must never execute on the {{Driver}} process.
> We saw an invocation of {{fatal}} deadlock in our internal CI since its invocation in {{StorageLocalResourceProviderProcess::sendResourceProviderStateUpdate}} wasn't explicitly {{defer}}'d, and by accident was executing on the {{Driver}}'s process.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)