[ 
https://issues.apache.org/jira/browse/FLINK-34573?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

LSZ updated FLINK-34573:
------------------------
    Description: 
we havae a flink job , jst one taskmanger;

when use high presure as soure data,it will be stuck. sometimes it will be run 
1d ,somtimes it will be run 30min.

!stuck.PNG!

like this: (13:30 the taskmanager reboot,then run 30min, result is stuck )

test 3 cases:

1: low presure (1200eps ),  it will run 30 min or 1d 。

2: close checkpoint , it will run 3d , high presure (1800eps) ,did not run 
stuck。

3:double the orignal  managermemory, it still stuck, jst The appearance time 
has been changed to 3 days from 30mins.

!rate.PNG!

 

the threads dump info ,when high presure , cpu 90%~100%:

[^tm-thread-dump-chk-0123[1].json]

this is the normal info, when the low presure :
[^tm-thread-dump-no-lock-0123[1].json]

 

 

  was:
we havae a flink job , jst one taskmanger;

when use high presure as soure data,it will be stuck. 

!stuck.PNG!

 

the threads dump info ,when high presure , cpu 90%~100%:

[^tm-thread-dump-chk-0123[1].json]

this is the normal info, when the low presure :
[^tm-thread-dump-no-lock-0123[1].json]

 

 


> the task is stuck on the high presure
> -------------------------------------
>
>                 Key: FLINK-34573
>                 URL: https://issues.apache.org/jira/browse/FLINK-34573
>             Project: Flink
>          Issue Type: Bug
>          Components: Runtime / Network
>    Affects Versions: 1.14.3
>            Reporter: LSZ
>            Priority: Blocker
>         Attachments: rate.PNG, stuck.PNG, tm-thread-dump-chk-0123[1].json, 
> tm-thread-dump-no-lock-0123[1].json
>
>   Original Estimate: 120h
>  Remaining Estimate: 120h
>
> we havae a flink job , jst one taskmanger;
> when use high presure as soure data,it will be stuck. sometimes it will be 
> run 1d ,somtimes it will be run 30min.
> !stuck.PNG!
> like this: (13:30 the taskmanager reboot,then run 30min, result is stuck )
> test 3 cases:
> 1: low presure (1200eps ),  it will run 30 min or 1d 。
> 2: close checkpoint , it will run 3d , high presure (1800eps) ,did not run 
> stuck。
> 3:double the orignal  managermemory, it still stuck, jst The appearance time 
> has been changed to 3 days from 30mins.
> !rate.PNG!
>  
> the threads dump info ,when high presure , cpu 90%~100%:
> [^tm-thread-dump-chk-0123[1].json]
> this is the normal info, when the low presure :
> [^tm-thread-dump-no-lock-0123[1].json]
>  
>  



--
This message was sent by Atlassian Jira
(v8.20.10#820010)

Reply via email to