osdir.com


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

答复: No data issued by flink window after a few hours


Hi vino,

I found when job run a period of time, the currentLowWatermark will be like a time far from now like 1539618594000 (2018/10/15 23:49:54), but now is 2018/10/15 14:29:00 .

I checked event time from data, there’re no event time far from now.

 

Yours,

September

 


发件人: 潘 功森 <pangongsen@xxxxxxxxxxx>
发送时间: Friday, October 12, 2018 2:05:33 PM
收件人: vino yang
抄送: user; dwysakowicz@xxxxxxxxxx
主题: 答复: No data issued by flink window after a few hours
 

Hi,

It stopped to update now.

 

Yours,

September

 


发件人: 潘 功森 <pangongsen@xxxxxxxxxxx>
发送时间: Friday, October 12, 2018 1:49:58 PM
收件人: vino yang
抄送: user; dwysakowicz@xxxxxxxxxx
主题: 答复: No data issued by flink window after a few hours
 

Hi,

       I chose the ten seconds and fifteen minutes windows, and it looks fine when running few minutes as below pictures.

 

 

But I found therere no changes  between 13:34 and 13:36, but it also could be issued , and I can tolerate it.

 

If no results issued after windows ends, I will take pictures again for you.

 

Yours,

September

 

发件人: vino yang
发送时间: 20181012 13:04
主题: Re: No data issued by flink window after a few hours

 

Hi gongsen,

 

Since you are running well locally, it should not be a configuration issue. 

You can refer to the Flink UI to see if your checkpoint is delayed. 

I hope that you can follow the instructions in the documentation[1] and provide some screenshots that will help the community help locate the problem.

 

Thanks, vino.

 

 

潘 功森 <pangongsen@xxxxxxxxxxx> 20181012日周五 上午11:10写道:

Please look at the mail below,  the others were out of data cause the bad network.

 

Yours,

September

 

发件人: 功森 <pangongsen@xxxxxxxxxxx>
发送时间: Friday, October 12, 2018 11:05:49 AM
收件人: vino yang
抄送: user
主题: 答复: No data issued by flink window after a few hours

 

Hi,

I found the pictures maybe too big and the net here not so good, so the mail I wrote is not sent sucsessfully last night.

Yes, I used event time.

I found watermarks fired normally when the job started, but it stopped and no changed after running hours.

And I changed as fs state backend, I configured at flink-conf.yaml below:

State.backend: filesystem

State.backend.fs.checkpointdir: file:///xxx/checkpoints

 

I found there are many checkpoints saved in the file.

But the watermarks also will be stoppped.

 

And therere two enviroments.

The local running well, and the problem occurred on site.

Therere two differences:

1.     Data amount very small in local and very huge on site.

2.     On site configurations:

Akka.ask.timeout: 2min

Akka.lookup.timeout: 2min

Akka.client.timeout: 3min

Akka.log.lifecycle.events: on

Akka.tcp.timeout: 60s

 

But not configured at local.

So therere any influences?

Please help me

     

 

Yours,

September

 

发件人: vino yang
发送时间: 20181011 14:56
收件人: pangongsen@xxxxxxxxxxx
抄送: user
主题: Re: No data issued by flink window after a few hours

 

Hi gongsen,

 

Have you used event time as time semantics? If so, then the possible problem is related to watermark. 

Since I don't know the details of your program, it's hard to make a conclusion. You can check if your watermark is firing normally.

 

Thanks, vino.

 

潘 功森 <pangongsen@xxxxxxxxxxx> 20181011日周四 下午12:12写道:

Hi

I changed as below configurationsand it looks fine when job started.

But therere no results issued when window ends after running about six hours, and no errors and exceptions.

How can I position the question?

 

Yours,

September

 

cid:image001.png@01D4621B.8346A0B0

发件人: 功森 <pangongsen@xxxxxxxxxxx>
发送时间: Wednesday, October 10, 2018 2:44:48 PM
收件人: vino yang
抄送: user
主题: 答复: No data issued by flink window after a few hours

 

Hi

 

Cause default state size in one hour is too smalland the max window size is 24 hours, so I used 500M.

 

MemoryStateBackend stateBackend = new MemoryStateBackend(MAX_STATE_SIZE);//500M
env.setStateBackend(stateBackend);

 

And I found Irrespective of the configured maximal state size, the state cannot be larger than the akka frame size.

So I add a config in flink-comf.yaml:

akka.framesize: 524288000b

 

What else do I have to pay attention to?

 

Yours,

September

 

cid:image001.png@01D4621B.8346A0B0

发件人: vino yang <yanghua1127@xxxxxxxxx>
发送时间: Wednesday, October 10, 2018 11:45:31 AM
收件人: pangongsen@xxxxxxxxxxx
抄送: user
主题: Re: No data issued by flink window after a few hours

 

Hi,

 

I saw the exception image you provided. Based on the exception message, it seems you used the default max state size (5MB).

 

You can specify the max state size to override the default value. Try :

 

MemoryStateBackend stateBackend = new MemoryStateBackend(theSizeOfBytes);

 

Please note that you need to reserve enough memory for Flink.

 

Thanks, vino.

 

潘 功森 <pangongsen@xxxxxxxxxxx> 20181010日周三 上午11:36写道:

Please have a look about my last mail.

 

When the cached window data is too large, how?

 

Yours,

September

 

cid:image002.png@01D4621B.8346A0B0

发件人: vino yang <yanghua1127@xxxxxxxxx>
发送时间: Wednesday, October 10, 2018 11:33:48 AM
收件人: pangongsen@xxxxxxxxxxx
抄送: user
主题: Re: No data issued by flink window after a few hours

 

Hi,

 

Did you mean "computer momery" referring to Memory Statebackend? 

The Flink window mechanism is internally based on State, and this is done for fault tolerance. 

If you introduce external storage, it will break its design and bring other problems.

 

Thanks, vino.

 

潘 功森 <pangongsen@xxxxxxxxxxx> 20181010日周三 上午11:02写道:

Hi,

"ram to cache the distinct data about sliding window" means I used computer momery not the third part db to cache the data need used in window.

the data need used in window means such as the sliding window is 1 hour, and I need to count the distinct users, I need to cache the user id about one hour.

Cause therere no related errors.

Yours,

September

 

cid:image003.png@01D4621B.8346A0B0

发件人: vino yang <yanghua1127@xxxxxxxxx>
发送时间: Wednesday, October 10, 2018 10:49:43 AM
抄送: user
主题: Re: No data issued by flink window after a few hours

 

Hi,

 

Can you explain what "ram to cache the distinct data about sliding window" mean? 

The information you provide is too small and will not help others to help you analyze the problem and provide advice.

 

In addition, regarding the usage of Flink related issues, please only send mail to the user mailing list. 

The dev mailing list is mainly used to discuss development related issues.

 

Thanks vino.

 

? ?? <pangongsen@xxxxxxxxxxx> 20181010日周三 上午10:37写道:

Hi all,
    I used flink window, and when the job begins, we could get the results of windiow.But there
re no results issued after a few hours.
    I found the job is still running and no errors, and the data not used window all can be issued.
    By the way, I used Flink 1.3.2 and ram to cache the distinct data about sliding window.

Yours,
    September