[
https://issues.apache.org/jira/browse/FLINK-11374?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16917602#comment-16917602
]
Till Rohrmann commented on FLINK-11374:
---------------------------------------
Filtering could also happen on the web UI side if we add the timestamps of the
exceptions, right? I would prefer this approach atm.
{{JobExceptionsHandler}} reports the failure cause for the overall job.
Moreover, we don't store the failure causes for previous job executions. Only
the very last root failure cause is stored. Adding this would be a bigger
change to the {{ExecutionGraph}} which I would like to avoid at the moment
because this structure might soon be replace with something different.
The limit of 20 exceptions are the exceptions from the different tasks
belonging to the latest execution. How would you deal with REST responses
becoming super large if you have many different subtasks? I think we need a
good answer for this before removing this limit.
> See more failover and can filter by time range
> ----------------------------------------------
>
> Key: FLINK-11374
> URL: https://issues.apache.org/jira/browse/FLINK-11374
> Project: Flink
> Issue Type: Improvement
> Components: Runtime / REST, Runtime / Web Frontend
> Reporter: lining
> Assignee: lining
> Priority: Major
> Attachments: image-2019-01-22-11-40-53-135.png,
> image-2019-01-22-11-42-33-808.png
>
>
> Now failover just show limit size task failover latest time. If task has
> failed many time, we can not see the earlier time failover. Can we add filter
> by time to see failover which contains task attemp fail msg.
--
This message was sent by Atlassian Jira
(v8.3.2#803003)
