[ https://issues.apache.org/jira/browse/FLINK-6043?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15940116#comment-15940116 ]
ASF GitHub Bot commented on FLINK-6043: --------------------------------------- Github user zentol commented on the issue: https://github.com/apache/flink/pull/3583 @tillrohrmann I've addressed your comments regarding ErrorInfo being volatile/fields being final as well as markTimestamp generating and returning the timestamp. @StephanEwen @tillrohrmann Regarding the timestamp in notifyStateTransiton etc. :The webUI displays 2 kind of exceptions: 1) the root exception that is stored in the ExecutionGraph and 2) a list of task exceptions that are stored in the Executions. IMO, if the exception stored in an Execution is also the root exception in the EG then the timestamps should match to be consistent and not give off the impression that these are unrelated/different. To that end the Execution has to supply the Exception as well as the timestamp. We could create the ErrorInfo directly in the Execution and pass that on instead of just the exception. This would make it move obvious that the TS is related to the exception i suppose. > Display time when exceptions/root cause of failure happened > ----------------------------------------------------------- > > Key: FLINK-6043 > URL: https://issues.apache.org/jira/browse/FLINK-6043 > Project: Flink > Issue Type: Improvement > Components: Webfrontend > Affects Versions: 1.3.0 > Reporter: Till Rohrmann > Assignee: Chesnay Schepler > Priority: Minor > > In order to better understand the behaviour of Flink jobs, it would be nice > to add timestamp information to exception causing the job to restart or to > fail. This information could then be displayed in the web UI making it easier > for the user to understand what happened when. -- This message was sent by Atlassian JIRA (v6.3.15#6346)