Hi,

we had quite some severe memory problems with Job DSL Plugin, but then, we were 
running it quite often (like, all the time...). In the end, it didn't even 
survive one working day.

Using the JavaMelody Monitoring Plugin, we saw an ever increasing sawtooth in 
the "Used non-heap memory" and the "loaded classes count" graphs. We increased 
max memory and PermGen, but that only delayed the inevitable, it would always 
hit the max after a while.

Running JProfiler on a test instance, we found two problems:
- JobParent gets injected into the Groovy compiler (only to output some 
deprecation warning). This by now has been removed in 1.36 
(https://github.com/jenkinsci/job-dsl-plugin/commit/066606f)
- In our dsl scripts, we stored quite a lot of information (stored as local 
variables of JobParent class). We had to null all these variables, otherwise 
their content was never gc'ed. Seems there is yet another leak of JobParent 
somewhere, but we did not find that.

We use monkey patching and Categories in our DSL scripts, but no mixins. Memory 
is quite stable now, but we still restart Jenkins nightly, mostly out of 
habit...

Marc

On 11 Sep 2015, at 04:37, Daniel Beck <m...@beckweb.net> wrote:

> 
> On 11.09.2015, at 01:51, Slide <slide.o....@gmail.com> wrote:
> 
>> I would also like to know if anyone can help me figure out how to profile 
>> Jenkins (while running under Tomcat) so I can determine where the memory 
>> issues are coming from.
> 
> Anything interesting in the heap dump? (Does /computer/(master)/heapDump work 
> when running in Tomcat?)
> 
> If you have a bunch of GroovyClassLoaders, could be JENKINS-23762.

-- 
You received this message because you are subscribed to the Google Groups 
"Jenkins Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to jenkinsci-users+unsubscr...@googlegroups.com.
To view this discussion on the web visit 
https://groups.google.com/d/msgid/jenkinsci-users/3535EFE8-98AE-400A-9A60-75C7670B5D27%40ebay.com.
For more options, visit https://groups.google.com/d/optout.

Reply via email to