[
https://issues.apache.org/jira/browse/OAK-2145?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14151615#comment-14151615
]
Alex Parvulescu commented on OAK-2145:
--------------------------------------
Could the fact that the FailoverClient doesn't close the _group_ and _executor_
on each run be the culprit?
In the beginning the client used to bootstrap and shutdown on the #run method,
but following this commit [0] it will only cleanup on #close(), but still
produce new _group_ and _executor_ instances on each run (each 5 seconds) which
will not get cleaned and I'm assuming will not release the open handlers.
So I think we should either: init only once (not on each #run call), or cleanup
after each run.
My preference is the latter, as per the original design.
[0]
https://github.com/apache/jackrabbit-oak/commit/3f8ba4216349b12d6424f8579901497412905282#diff-8675625fa2d4f154c0a71b0a976376bfL106
> TarMK cold standby: file handle leak
> ------------------------------------
>
> Key: OAK-2145
> URL: https://issues.apache.org/jira/browse/OAK-2145
> Project: Jackrabbit Oak
> Issue Type: Bug
> Components: segmentmk
> Reporter: Manfred Baedke
>
> The TarMK cold standby client apparently doesn't close all used file streams
> properly. When synchronizing very huge repositories the OS may run out of
> file handles.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
