[ 
https://issues.apache.org/jira/browse/KAFKA-2252?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14586744#comment-14586744
 ] 

Gwen Shapira commented on KAFKA-2252:
-------------------------------------

Created reviewboard https://reviews.apache.org/r/35474/diff/
 against branch trunk

> Socket connection closing is logged, but not corresponding opening of socket
> ----------------------------------------------------------------------------
>
>                 Key: KAFKA-2252
>                 URL: https://issues.apache.org/jira/browse/KAFKA-2252
>             Project: Kafka
>          Issue Type: Bug
>            Reporter: Jason Rosenberg
>         Attachments: KAFKA-2252.patch
>
>
> (using 0.8.2.1)
> We see a large number of "Closing socket connection" logging to the broker 
> logs, e.g.:
> {code}
> 2015-06-04 16:49:30,262  INFO [kafka-network-thread-27330-2] 
> network.Processor - Closing socket connection to /1.2.3.4.
> 2015-06-04 16:49:30,262  INFO [kafka-network-thread-27330-0] 
> network.Processor - Closing socket connection to /5.6.7.8.
> 2015-06-04 16:49:30,695  INFO [kafka-network-thread-27330-0] 
> network.Processor - Closing socket connection to /9.10.11.12.
> 2015-06-04 16:49:31,465  INFO [kafka-network-thread-27330-1] 
> network.Processor - Closing socket connection to /13.14.15.16.
> 2015-06-04 16:49:31,806  INFO [kafka-network-thread-27330-0] 
> network.Processor - Closing socket connection to /17.18.19.20.
> 2015-06-04 16:49:31,842  INFO [kafka-network-thread-27330-2] 
> network.Processor - Closing socket connection to /21.22.23.24.
> {code}
> However, we have no corresponding logging for when these connections are 
> established.  Consequently, it's not very useful to see a flood of closed 
> connections, etc.  I'd think we'd want to see the corresponding 'connection 
> established' messages, also logged as INFO.
> Occasionally, we see a flood of the above messages, and have no idea as to 
> whether it indicates a problem, etc.  (Sometimes it might be due to an 
> ongoing rolling restart, or a change in the Zookeeper cluster).



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Reply via email to