[
https://issues.apache.org/jira/browse/YARN-8895?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16667544#comment-16667544
]
Young Chen commented on YARN-8895:
----------------------------------
Hi [~leftnoteasy] -
1 - I agree that we should make this a new field and leave diagnostics as is
until we have equivalent or better functionality with the structured errors.
2 - The most important changes will be:
* code in NM and the RM to construct these structured errors as close to the
source as possible
* Protobuf changes for the NM to communicate these errors to the RM, maybe in
container status reports if the exit was abnormal
* Update RM failover to save these structured errors during restarts
As for details, I'm still debating whether a pluggable structured error
implementation would be worth it - I think error structures are more or less
very similar. Error code, message, description, source component, user/system,
etc..
What do you think?
> Improve YARN Error diagnostics
> -------------------------------
>
> Key: YARN-8895
> URL: https://issues.apache.org/jira/browse/YARN-8895
> Project: Hadoop YARN
> Issue Type: Improvement
> Reporter: Young Chen
> Assignee: Young Chen
> Priority: Minor
>
> Currently identifying error sources can be quite difficult, as they are
> written into an unstructured string "diagnostics" field. This is present in
> container statuses returned to the RM and in application attempts in the RM.
> These errors are difficult to classify without hard-coding diagnostic string
> searches.
> This Jira aims to add a structured error field in NM and RM that preserves
> failure information and source component to enable faster and clearer error
> diagnosis
>
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: yarn-issues-unsubscr...@hadoop.apache.org
For additional commands, e-mail: yarn-issues-h...@hadoop.apache.org
