[
https://issues.apache.org/jira/browse/TEZ-3029?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15140257#comment-15140257
]
Siddharth Seth commented on TEZ-3029:
-------------------------------------
bq. dag names are not unique. On error should not use it as an identifier.
Thank you for the reminder. Will fix this.
bq. non-needed change in ContainerLauncher.java?
Will remove the blank lines.
bq. s/diag= "/diagnostics="/ - better for readability.
Will fix in the next patch.
bq. why is ServicePluginError needed? Doesnt the fatal error reporting handle
the permanent case? UPDATE: Looks like onError is being removed - should tag
this jira as an incompatible change?
It is incompatible - on unstable APIs though. I'll put it in the incompatible
section though.
bq. processNonFatalServiceErrorReport() - shouldnt this fail the dag instead of
killing it?
I had marked the dags as killed because the DAG itself has not necessarily
failed. If I were to write a user plugin to determine the state of the DAG
based on these service errors - the state would be KILLED. [~prasanth_j] - any
thoughts on this ? [~hitesh] - reasoning for FAIL instead of KILL ?
> Add an onError method to service plugin contexts
> ------------------------------------------------
>
> Key: TEZ-3029
> URL: https://issues.apache.org/jira/browse/TEZ-3029
> Project: Apache Tez
> Issue Type: Sub-task
> Reporter: Siddharth Seth
> Assignee: Siddharth Seth
> Attachments: TEZ-3029.1.txt
>
>
> This is to indicate errors which may occur while running threads etc.
> One bit to be careful about is that this introduces two means of reporting
> errors - which has been challenging to handle in the past for
> Input/Processor/OutputContext - so whether to do this or not needs to be
> thought through.
> TaskSchedulerContext already has this method.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
