[
https://issues.apache.org/jira/browse/NIFI-9012?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Michal Šunka updated NIFI-9012:
-------------------------------
Attachment: thread-dump2.txt
thread-dump.txt
> InvokeHttps
> -----------
>
> Key: NIFI-9012
> URL: https://issues.apache.org/jira/browse/NIFI-9012
> Project: Apache NiFi
> Issue Type: Bug
> Affects Versions: 1.14.0
> Environment: Deckerized NiFi with container from Docker Hub, no
> custom processors
> Reporter: Michal Šunka
> Priority: Major
> Attachments: thread-dump.txt, thread-dump2.txt
>
>
> I was running a NiFi version 1.12.1 without any problems with this regard.
> Yesterday I updated to version 1.14.0 (really wanted to get fix
> https://issues.apache.org/jira/browse/NIFI-7849). But now I am in a way worse
> state.
> I use InvokeHTTP processor to self-connect to NiFi API to get count of
> flowfiles being processed by given processor group.
> However this InvokeHTTP started getting stuck. Processor would just spin up a
> new thread and this will sit there indefinitely (six hours at-least),
> blocking incoming flowfile. After forceful termination of processor (simple
> stop is not enough) the flowfile would get released and next start will
> consume the flowfile just fine. (Timeouts are: connection 5 seconds, read 10
> seconds and idle 1 minute.) And of course log is clear of any errors which
> seems to be connected to this...
> I had noticed similar issue also with other processors, namely MergeContent
> processor or ExecuteSQLProcessor - e.g. flowfile gets stuck for 15 minutes
> (merge bin max age is 5 seconds). After forced termination all starts working
> (for some time - a few dozens of minutes).
> Oh and in all cases the thread hangs in terminating state (red thread icon
> with a digit in parenthesis; the digit sits there all the time and wont
> disappear). Only restart of whole NiFi clears the issue. For some time.
>
> Maybe some disk I/O issue?
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
