[
https://issues.apache.org/jira/browse/SYNCOPE-1763?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Francesco Chicchiriccò updated SYNCOPE-1763:
--------------------------------------------
Component/s: client
(was: console)
(was: enduser)
(was: sra)
(was: wa)
Description:
Just after the upgrade of CXF from 3.5.6 to 3.6.0 (commit
[2252d74209bb944d90dd20579c2468597e4f79f3|https://github.com/apache/syncope/commit/2252d74209bb944d90dd20579c2468597e4f79f3])
the number of open FDs is in constant increase.
This behaviour occours everytime an REST request is performed against the Core
component.
This issue has been discovered investigating a "Too many open files" exception
raised by WA component during delete expiring gauth tokens job execution.
Generally, considering an execution of this jov every 35 seconds, after 16
hours, more or less, WA becomes completely unavailable cause this issue.
The issue has been reproduced in the embedded env and analized by lsof tool.
This is an example before the commit above
{code:java}
$ while [ 1 ]; do sudo lsof -u fabio 2> /dev/null | grep 54201| grep pipe | wc
-l; sleep 60; done 17 17
17
17
17{code}
and the following, the behaviour just after
{code:java}
$ while [ 1 ]; do sudo lsof -u fabio 2> /dev/null | grep 45581| grep pipe | wc
-l; sleep 60; done
157
179
183
185
189{code}
As you can see the difference is evident.
was:
Just after the upgrade of CXF from 3.5.6 to 3.6.0 (commit
2252d74209bb944d90dd20579c2468597e4f79f3) opened FDs increase in number without
ever decreasing.
This behaviour occour evrytime a cxf request is performed towards the Core
component.
This issue has been discovered investigating a "Too many open files" exception
raised by WA component during delete expiring gauth tokens job execution.
Generally, considering an execution of this jov every 35 seconds, after 16
hours, more or less, WA becomes completely unavailable cause this issue.
The issue has been reproduced in the embedded env and analized by lsof tool.
This is an example before the commit above
{code:java}
$ while [ 1 ]; do sudo lsof -u fabio 2> /dev/null | grep 54201| grep pipe | wc
-l; sleep 60; done 17 17
17
17
17{code}
and the following, the behaviour just after
{code:java}
$ while [ 1 ]; do sudo lsof -u fabio 2> /dev/null | grep 45581| grep pipe | wc
-l; sleep 60; done
157
179
183
185
189{code}
As you can see the difference is evident.
Issue Type: Bug (was: Improvement)
Summary: Constant increase of open files after upgrade to CXF 3.6.0
(was: CXF 3.6.0 cause increasing of opened files without ever decrease)
> Constant increase of open files after upgrade to CXF 3.6.0
> ----------------------------------------------------------
>
> Key: SYNCOPE-1763
> URL: https://issues.apache.org/jira/browse/SYNCOPE-1763
> Project: Syncope
> Issue Type: Bug
> Components: client
> Affects Versions: 3.0.4
> Reporter: Fabio Martelli
> Priority: Major
> Fix For: 3.0.4, 4.0.0
>
>
> Just after the upgrade of CXF from 3.5.6 to 3.6.0 (commit
> [2252d74209bb944d90dd20579c2468597e4f79f3|https://github.com/apache/syncope/commit/2252d74209bb944d90dd20579c2468597e4f79f3])
> the number of open FDs is in constant increase.
> This behaviour occours everytime an REST request is performed against the
> Core component.
> This issue has been discovered investigating a "Too many open files"
> exception raised by WA component during delete expiring gauth tokens job
> execution. Generally, considering an execution of this jov every 35 seconds,
> after 16 hours, more or less, WA becomes completely unavailable cause this
> issue.
> The issue has been reproduced in the embedded env and analized by lsof tool.
> This is an example before the commit above
> {code:java}
> $ while [ 1 ]; do sudo lsof -u fabio 2> /dev/null | grep 54201| grep pipe |
> wc -l; sleep 60; done 17 17
> 17
> 17
> 17{code}
> and the following, the behaviour just after
> {code:java}
> $ while [ 1 ]; do sudo lsof -u fabio 2> /dev/null | grep 45581| grep pipe |
> wc -l; sleep 60; done
> 157
> 179
> 183
> 185
> 189{code}
> As you can see the difference is evident.
>
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
