You are viewing a plain text version of this content. The canonical link for it is here.
Posted to yarn-issues@hadoop.apache.org by "Szilard Nemeth (Jira)" <ji...@apache.org> on 2020/10/22 07:58:00 UTC
[jira] [Commented] (YARN-10323) [Umbrella] YARN Diagnostic
collector
[ https://issues.apache.org/jira/browse/YARN-10323?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17218842#comment-17218842 ]
Szilard Nemeth commented on YARN-10323:
---------------------------------------
Thanks [~bteke] for making this clear, I quickly prepared this Jira and completely forgot to mention that the design document is written by [~prabhujoseph]. We just discussed I will create this Jira itself. Didn't want to take credits or something similar. Apologies for this.
> [Umbrella] YARN Diagnostic collector
> ------------------------------------
>
> Key: YARN-10323
> URL: https://issues.apache.org/jira/browse/YARN-10323
> Project: Hadoop YARN
> Issue Type: Improvement
> Reporter: Szilard Nemeth
> Assignee: Benjamin Teke
> Priority: Major
> Attachments: YARN-10323 Design of Diagnostics Collector.pdf
>
>
> Troubleshooting YARN problems can be difficult on a production environment.
> Collecting data before problems occur or actively collecting data in an on-demand basis could truly help tracking down issues.
> Some examples:
> 1. If application is hanging, application logs along with RM / NM logs could be collected, plus jstack of either the YARN daemons or the application container.
> 2. Similarly, when an application fails we may collect data.
> 3. Scheduler issues are quite common so good tooling that helps to spot issues would be crucial.
> Design document in the attachments section.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: yarn-issues-unsubscribe@hadoop.apache.org
For additional commands, e-mail: yarn-issues-help@hadoop.apache.org