There were a lot of connections which state is “CLOSE_WAIT” appeared when a few days later after flink jobManager is started.
The process name of pid 20180 is 'flink jobManager'. The stack trace is blow: |
Hi XiangWei,
this could be a resource leak, i.e. a socket not getting closed, but I was unable to reproduce that behaviour. Maybe Chesnay (cc'd) has an idea on how/ where this may happen. Can you tell us a bit more on what you where doing / how the webinterface was used? Is there a way to reproduce the behaviour? Nico PS: please don't post text output as images - this makes reading/finding the email and following the thread much more complicated. Please copy the text instead. On Thursday, 29 June 2017 06:36:47 CEST XiangWei Huang wrote: > There were a lot of connections which state is “CLOSE_WAIT” appeared when > a few days later after flink jobManager is started. > > The process name of pid 20180 is 'flink jobManager'. > The stack trace is blow: signature.asc (201 bytes) Download Attachment |
Hello,
is this list constantly growing? If you reload the WebUI do they pile up again? My guess would be the watermarks display is overloading the metrics handler. If i remember correctly the WebUI keeps fetching the watermark metrics regardless of what page you're looking at. Things to investigate would be whether it continues to query for watermarks for finished jobs, and whether requests are batched across tasks & jobs. Regards, Chesnay On 30.06.2017 18:05, Nico Kruber wrote: > Hi XiangWei, > this could be a resource leak, i.e. a socket not getting closed, but I was > unable to reproduce that behaviour. Maybe Chesnay (cc'd) has an idea on how/ > where this may happen. > > Can you tell us a bit more on what you where doing / how the webinterface was > used? Is there a way to reproduce the behaviour? > > > > Nico > > > PS: please don't post text output as images - this makes reading/finding the > email and following the thread much more complicated. Please copy the text > instead. > > On Thursday, 29 June 2017 06:36:47 CEST XiangWei Huang wrote: >> There were a lot of connections which state is “CLOSE_WAIT” appeared when >> a few days later after flink jobManager is started. >> >> The process name of pid 20180 is 'flink jobManager'. >> The stack trace is blow: |
hi,
Sorry for replying so late. I have met this issue again and the list is constantly keep growing even if i close the page ,until the website is been unavailable. This issue appeared each time i add metrics for a job from web ui. by the way ,the version of Flink is 1.3.1 Regards, XiangWei |
Hello,
Could you tell us which browser you are using, including the version? (and maybe try out if the issue persists with a different one) Regards, Chesnay On 25.07.2017 05:20, XiangWei Huang wrote: > hi, > > Sorry for replying so late. > I have met this issue again and the list is constantly keep growing even if > i close the page ,until the website is been unavailable. > > This issue appeared each time i add metrics for a job from web ui. > > by the way ,the version of Flink is 1.3.1 > > > > Regards, > XiangWei > > > > > -- > View this message in context: http://apache-flink-user-mailing-list-archive.2336050.n4.nabble.com/a-lot-of-connections-in-state-CLOSE-WAIT-tp14046p14422.html > Sent from the Apache Flink User Mailing List archive. mailing list archive at Nabble.com. > |
hi,
The browser i am using is Google Chrome with version 59.0.3071.115 and the issue persists when i tried Firefox. Regards, XiangWei > 在 2017年7月25日,17:48,Chesnay Schepler <[hidden email]> 写道: > > Hello, > > Could you tell us which browser you are using, including the version? > (and maybe try out if the issue persists with a different one) > > Regards, > Chesnay > > On 25.07.2017 05:20, XiangWei Huang wrote: >> hi, >> >> Sorry for replying so late. >> I have met this issue again and the list is constantly keep growing even if >> i close the page ,until the website is been unavailable. >> >> This issue appeared each time i add metrics for a job from web ui. >> >> by the way ,the version of Flink is 1.3.1 >> >> >> >> Regards, >> XiangWei >> >> >> >> >> -- >> View this message in context: http://apache-flink-user-mailing-list-archive.2336050.n4.nabble.com/a-lot-of-connections-in-state-CLOSE-WAIT-tp14046p14422.html >> Sent from the Apache Flink User Mailing List archive. mailing list archive at Nabble.com. >> > |
So this only happens when you
select a metric? Without a selected metric everything works fine?
Are the metrics you selected shown correctly? Did you modify the "jobmanager.web.refresh-interval" setting? (maybe check the flink-conf-yaml for the current setting) On 26.07.2017 04:57, XiangWei Huang wrote: hi, The browser i am using is Google Chrome with version 59.0.3071.115 and the issue persists when i tried Firefox. Regards, XiangWei在 2017年7月25日,17:48,Chesnay Schepler [hidden email] 写道: Hello, Could you tell us which browser you are using, including the version? (and maybe try out if the issue persists with a different one) Regards, Chesnay On 25.07.2017 05:20, XiangWei Huang wrote:hi, Sorry for replying so late. I have met this issue again and the list is constantly keep growing even if i close the page ,until the website is been unavailable. This issue appeared each time i add metrics for a job from web ui. by the way ,the version of Flink is 1.3.1 Regards, XiangWei -- View this message in context: http://apache-flink-user-mailing-list-archive.2336050.n4.nabble.com/a-lot-of-connections-in-state-CLOSE-WAIT-tp14046p14422.html Sent from the Apache Flink User Mailing List archive. mailing list archive at Nabble.com.
|
1. yes and yes.
2. Yes,it was shown correctly. 3.I wasn’t modify this setting.
|
FLINK-7368 may be the reason for this
behavior.
On 31.07.2017 03:54, XiangWei Huang wrote: 1. yes and yes.
|
It seems so,thanks for your reply Chesnay.
|
Free forum by Nabble | Edit this page |