ElasticSearch Checkpointing taking too much time

[shashank734]

I am using flink 1.5.3, In this i am using elastic search sink. In this
checkpoints and savepoints are failing, I have already given 50 minutes
timeouts. After looking into details only elastic search sink checkpoints
are taking time 30-35 mins. But state size and buffer size is 0 in that.
Don't know why it's taking too much time when it's state size is 0.



--
Sent from: http://apache-flink-user-mailing-list-archive.2336050.n4.nabble.com/

[shashank734]

Update:

I am using parallelism 1 on this... is this issue?



--
Sent from: http://apache-flink-user-mailing-list-archive.2336050.n4.nabble.com/

[Hequn Cheng]

Hi shashank,

The parallelism won't be the problem.

Did the checkpoint succeed finally? I think it may be that the data processing is blocked so that the checkpoint can not been successful. You can check if there are any error logs in the TaskManager or jstack the taskmanager to see what's wrong the the task.

Best, Hequn

On Tue, Sep 11, 2018 at 10:22 PM shashank734 <[hidden email]> wrote:

Update:

I am using parallelism 1 on this... is this issue?



--
Sent from: http://apache-flink-user-mailing-list-archive.2336050.n4.nabble.com/

[vino yang]

Hi shashank,

Hequn's solution is right. In addition, what type of statebackend you use, please make sure that JM/TM can access related systems (such as HDFS). 

If you still can't locate the problem, you can set the log level to DEBUG and share your log information.

Thanks, vino.

Hequn Cheng <[hidden email]> 于2018年9月12日周三 上午9:22写道：

Hi shashank,

The parallelism won't be the problem.

Did the checkpoint succeed finally? I think it may be that the data processing is blocked so that the checkpoint can not been successful. You can check if there are any error logs in the TaskManager or jstack the taskmanager to see what's wrong the the task.

Best, Hequn

On Tue, Sep 11, 2018 at 10:22 PM shashank734 <[hidden email]> wrote:

Update:

I am using parallelism 1 on this... is this issue?



--
Sent from: http://apache-flink-user-mailing-list-archive.2336050.n4.nabble.com/

[shashank734]

Hi Hequn,  

Actually there are no error logs and to turn on debug mode I have to restart
the app, Actually, I am using around 25-30 operators all others are
completing successfully in less time only elastic search sink is taking too
much time. I am using around 6 Elastic search sinks all are taking time
around 25-30 mins. I have set 50 mins as timeout so 1-2 elastic search sinks
take more than that so checkpoint and savepoint fail with the timeout error.

I'll check jstack.

Thanks





--
Sent from: http://apache-flink-user-mailing-list-archive.2336050.n4.nabble.com/

[shashank734]

Hi, vino,

I have tried bot HDFS and filesystem and other checkpoints completed
successfully so access is not the issue. For debug mode, I have to restart
the app. I'll check and let you know thanks



--
Sent from: http://apache-flink-user-mailing-list-archive.2336050.n4.nabble.com/

[shashank734]

<http://apache-flink-user-mailing-list-archive.2336050.n4.nabble.com/file/t986/Screen_Shot_2018-09-12_at_12.png>



--
Sent from: http://apache-flink-user-mailing-list-archive.2336050.n4.nabble.com/