[jira] [Commented] (FLINK-24395) Checkpoint trigger time difference between log statement and web frontend

["Piotr Nowojski (Jira)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/FLINK-24395?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17430055#comment-17430055 ] 

Piotr Nowojski commented on FLINK-24395:
----------------------------------------

I'm not sure if this is an issue or not? Note that the timestamp that is later used in the webUI is present in the logged message {{1632823660151}}. So the one minor issue is that this timestamp is not pretty printed. 

The other issue, why this log message is printed with a delay is that we can log it only once we have already obtained checkpointId. Which happens in the ioExecutor thread. This probably explains why you have observed this couple of minute delay. But is this delay an actual problem here?

> Checkpoint trigger time difference between log statement and web frontend
> -------------------------------------------------------------------------
>
>                 Key: FLINK-24395
>                 URL: https://issues.apache.org/jira/browse/FLINK-24395
>             Project: Flink
>          Issue Type: Bug
>          Components: Runtime / Checkpointing
>    Affects Versions: 1.14.0
>            Reporter: Robert Metzger
>            Priority: Major
>         Attachments: image-2021-09-28-12-20-34-332.png
>
>
> Consider this checkpoint (68)
> {code}
> 2021-09-28 10:14:43,644 INFO  org.apache.flink.runtime.checkpoint.CheckpointCoordinator    [] - Triggering checkpoint 68 (type=CHECKPOINT) @ 1632823660151 for job 00000000000000000000000000000000.
> 2021-09-28 10:16:41,428 INFO  org.apache.flink.runtime.checkpoint.CheckpointCoordinator    [] - Completed checkpoint 68 for job 00000000000000000000000000000000 (128940015376 bytes, checkpointDuration=540908 ms, finalizationTime=369 ms).
> {code}
> And what is shown in the UI about it:
>  !image-2021-09-28-12-20-34-332.png! 
> The trigger time is off by ~7 minutes (the difference in the hours are timezone related). It seems that the trigger message is logged too late.
> (note that this has happened in a system where savepoint disposal is very slow)



--
This message was sent by Atlassian Jira
(v8.3.4#803005)