[
https://issues.apache.org/jira/browse/FLINK-17012?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17077150#comment-17077150
]
Piotr Nowojski commented on FLINK-17012:
----------------------------------------
Having separate metric for that seems a bit overkill. Especially that in the
long run, (Flink 2.0 after dropping DataSet API), we could unify StreamTask and
Task class into one and fix the task status to reflect state initialisation and
things like that. Once this is cleaned up, this metric would not be needed
anymore.
So maybe worth considering is a hacky solution to repurpose some existing
metric? Like not idle time set to -1 if it’s has not been started being
calculated yet? This -1 value during the initialisation might not be even
advertised as part of the public API.
> Expose stage of task initialization
> -----------------------------------
>
> Key: FLINK-17012
> URL: https://issues.apache.org/jira/browse/FLINK-17012
> Project: Flink
> Issue Type: Improvement
> Components: Runtime / Metrics
> Reporter: Wenlong Lyu
> Priority: Major
>
> Currently a task switches to running before fully initialized, does not take
> state initialization and operator initialization(#open ) in to account, which
> may take long time to finish. As a result, there would be a weird phenomenon
> that all tasks are running but throughput is 0.
> I think it could be good if we can expose the initialization stage of tasks.
> What to you think?
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
