[ 
https://issues.apache.org/jira/browse/NIFI-7200?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17044915#comment-17044915
 ] 

Joe Witt commented on NIFI-7200:
--------------------------------

https://issues.apache.org/jira/browse/NIFI-7114?focusedCommentId=17037190&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-17037190

We are not seeing the same issue with "pipe" as reported above. While I can't 
post our exact lsof output, here's basically what it looks like for us:

...  IPv4   ... ... TCP <url1>:<port> -> <url2>:pcsyn-http (CLOSE_WAIT)

This line above, with the same url1 and url2, repeats 17k times! And it keeps 
growing larger the longer we keep the Nifi running. There are many many ports. 
It seems that these connections are not being closed. I can't find anything 
obvious on these site-to-site connections to determine what's going on, other 
than to say we have a very large volume of data going through this site-to-site 
- but not anywhere near 17,000 flowfiles in one moment.

We currently need to have a cron job that checks if our open files are 
approaching our threshold, and the cron job will restart the nifi in that case. 
Otherwise we can come back after a weekend (or even overnight!) to several dead 
nifis.


https://issues.apache.org/jira/browse/NIFI-7114?focusedCommentId=17037236&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-17037236


Joe Witt We use RAW connections with TLS. In the past hour, the number of open 
files that meet the above pattern increased from 17,409 to 18,540. Again, this 
is only for the open files that meet the pattern I posted earlier! And since we 
have several site to sites, you can see how this problem is compounded, with 
the different site to sites increasing open files.

We did not have this issue when we were on Nifi 1.9.2. It happened after we 
upgraded to 1.11.0 and since then our Nifis with a lot of site to sites have 
this issue.

Thanks for your help investigating this problem!



> IPv4 socket resource leak
> -------------------------
>
>                 Key: NIFI-7200
>                 URL: https://issues.apache.org/jira/browse/NIFI-7200
>             Project: Apache NiFi
>          Issue Type: Bug
>            Reporter: Joe Witt
>            Priority: Critical
>
> https://issues.apache.org/jira/browse/NIFI-7114?focusedCommentId=17044888&page=com.atlassian.jira.plugin.system.issuetabpanels%3Acomment-tabpanel#comment-17044888



--
This message was sent by Atlassian Jira
(v8.3.4#803005)

Reply via email to