Nice! +1
Em sex, 8 de mar de 2019 às 05:07, j4fm
escreveu:
> Hi Jon
>
> Would you be able to check and merge PR 438 for me? I've broken the work
> down into bite sizes. This first one is just to get Cxf/MP Rest client
> back
> in (all tests succeed and without slowness). Later I will submit s
Hi Jon
Would you be able to check and merge PR 438 for me? I've broken the work
down into bite sizes. This first one is just to get Cxf/MP Rest client back
in (all tests succeed and without slowness). Later I will submit some work
around TCK+testng to improve memory handling when running a larg
This is fixed now. PR on it's way, it was mostly just new exclusions needed
adding, duh.
--
Sent from: http://tomee-openejb.979440.n4.nabble.com/TomEE-Dev-f982480.html
When running tests in full here are the tests that appear to cause a leak...
just to see if anything common between them is obvious from your experience?
Just to be sure, the leak exceptions from these do not appear in the logs at
all with CXF 3.2.7.
org.jboss.cdi.tck.tests.extensions.configurat
Thanks for the detail. I'm slammed myself, but hopefully will get some time
to help tomorrow.
I'd encourage others to dig in if they can / wish to. A starting point
could be: try pulling out one of the CDI TCK apps, deploy it manually, and
then undeploy, and see what's not getting freed and what i
Thanks for merging. Buildbot is succeeding again.
So my testng changes help speed up the tests and improve the heap usage
(worth adding the changes in at some point) but it didn't solve the root
cause. I see these in the logs:
SEVERE [http-nio-37745-exec-211]
org.apache.catalina.loader.WebappCl
Shoot over a PR. I would definitely be interested in a little bit of detail
before merging so we all understand what's happening.
Cheers
Jon
On Mon, Mar 4, 2019 at 3:10 AM j4fm wrote:
> So I have a fix for the cdi-tomee tck which means we can add cxf 3.3.0 back
> in. I just want to wait for f
So I have a fix for the cdi-tomee tck which means we can add cxf 3.3.0 back
in. I just want to wait for full tests to complete before explaining in
detail but to sum it up the bug was in testng which was keeping hold of
threads.
--
Sent from: http://tomee-openejb.979440.n4.nabble.com/TomEE-Dev-
Hiya, would you be able to merge PR 430 to allow buildbot to pass again for
me?
Thanks
--
Sent from: http://tomee-openejb.979440.n4.nabble.com/TomEE-Dev-f982480.html
I agree. It would hide the problem
Le dim. 3 mars 2019 à 03:23, j4fm a écrit :
> I ran it again with 768MB extra head and it crashed right near the end. I
> expect 1GB or 1.5GB would definitely let it finish but it must not be
> releasing memory on each application undeployment. I don't think
I ran it again with 768MB extra head and it crashed right near the end. I
expect 1GB or 1.5GB would definitely let it finish but it must not be
releasing memory on each application undeployment. I don't think just
increasing the memory and covering up an underlying problem is a good idea.
--
S
I ran the :cdi-tomee tck to successful completion with an extra 756MB added
to the max heap for this arquillian run. I'm running it again just to make
sure. Would updating master with the config be acceptable? We could have
cxf 3.3.0+mp rest client back in if so.
--
Sent from: http://tomee-op
Hey Jon, do you know if your test run with an increased java heap froze up or
succeeded?
Thanks
--
Sent from: http://tomee-openejb.979440.n4.nabble.com/TomEE-Dev-f982480.html
Slowness is fixed, so buildbot is happier but not quite happy. Just for
awareness to avoid someone else looking at it too...
1. I forgot to update the CXF version in the examples\polling-parent. So
polling-web failed.
2. Also, it turns out CXF didn't add support for mp rest client 1.2.0 until
3
nice!
Em sáb, 2 de mar de 2019 às 06:47, Jonathan Gallimore <
jonathan.gallim...@gmail.com> escreveu:
> Done
>
> On Sat, 2 Mar 2019, 09:37 j4fm, wrote:
>
> > CXF 3.3.1 is also causing the problem.
> >
> > I've submitted PR 429 just to make buildbot happy again for the time
> being.
> > Would you
Done
On Sat, 2 Mar 2019, 09:37 j4fm, wrote:
> CXF 3.3.1 is also causing the problem.
>
> I've submitted PR 429 just to make buildbot happy again for the time being.
> Would you mind merging just to get the builds going again? Will be good to
> confirm it's happy with the other PRs.
>
> Once a r
CXF 3.3.1 is also causing the problem.
I've submitted PR 429 just to make buildbot happy again for the time being.
Would you mind merging just to get the builds going again? Will be good to
confirm it's happy with the other PRs.
Once a root cause/solution is found can bump CXF back up again.
I was seeing the same, but didn't get to dig into it deeper yet - its
unlikely I'll be able to do much over the weekend, but we'll see.
I did bump up the heap on my build to 6G (from 1.5) to see how far it gets.
Jon
On Sat, 2 Mar 2019, 08:54 j4fm, wrote:
> It's just changing CXF version to 3.3
It's just changing CXF version to 3.3 on it's own that causes it. Reverting
back to 3.2.7 solves it.
All the JAXB 2.8.2 and related dependency work seems to be fine.
Though I see the CXF mp client version bump adds in activation-api which
needs excluding.
I'm testing 3.3.1 out now just to check (
Did anyone take a look at the heap dump (I can make it available if
needed)? If you did, what did you find?
I left visualvm computing sizes when I went to bed last night. Not sure
where it's at. Hoping to get a look over the weekend.
Jon
On Sat, 2 Mar 2019, 07:27 j4fm, wrote:
> Building is no
What is the mvn command which you are using to run it?
Em sáb, 2 de mar de 2019 às 04:27, j4fm
escreveu:
> Building is not a problem, it's the CDI TCK test run causing out of memory
> exception in tomee
>
>
>
> --
> Sent from:
> http://tomee-openejb.979440.n4.nabble.com/TomEE-Dev-f982480.html
>
Building is not a problem, it's the CDI TCK test run causing out of memory
exception in tomee
--
Sent from: http://tomee-openejb.979440.n4.nabble.com/TomEE-Dev-f982480.html
Hi folks,
it is happening in a specif module?
I'm building tomee without samples and seems fine to me.
mvn clean install -Pquick
Em sáb, 2 de mar de 2019 às 00:36, j4fm
escreveu:
> Hey Jon, thanks for the test. Do you still have all my PRs in there after
> just removing the PR409?
>
> Yeah, I
Hey Jon, thanks for the test. Do you still have all my PRs in there after
just removing the PR409?
Yeah, I was looking around all the dependency changes that CXF (and JAXB
which it depends on) brings in.
I do have something I am working on, instead of excluding the older javax
conflicting depende
Thanks for the update James.
I reverted locally: https://github.com/apache/tomee/pull/409 , build tome
without tests and then:
mvn -pl tck/cdi-tomee --fail-at-end
-Djava.io.tmpdir=/tmp/tomee-trunk-ubuntu-jvm8 clean install -Pall-adapters
-Dsurefire.useFile=false -DdisableXmlReport=true
-Dopenejb.
False fix, it just got a lot further before an oom exception.
--
Sent from: http://tomee-openejb.979440.n4.nabble.com/TomEE-Dev-f982480.html
I have a fix for the slow build (works locally without memory error now). If
somebody wouldn't mind stopping the current buildbot task and I will create
a small PR.
Thanks in advance
--
Sent from: http://tomee-openejb.979440.n4.nabble.com/TomEE-Dev-f982480.html
Not sure it's ever going to finish. Seems to be grinding to a halt. You're
right, no point letting it run.
--
Sent from: http://tomee-openejb.979440.n4.nabble.com/TomEE-Dev-f982480.html
Would like to get the stats at the end of the CI build if possible too.
PR409 was the first merge to start the issue but there are no code changes
in there except to examples and the bumps of CXF and JAXB.
Do you see it just being cdi-tomee that's slowed up? I'm checking out it's
(indirect) depe
Waiting on my machine to crunch some numbers, but here's the summary:
Total Bytes: 2,726,889,443
Total Classes: 124,138
Total Instances: 54,413,232
Classloaders: 790
GC Roots: 3,842
Number of Objects Pending for Finalization: 0
That seems like a lot of cl
On the plus side, I have a heap dump here, so we can take a look.
Jon
On Fri, Mar 1, 2019 at 9:28 PM Jonathan Gallimore <
jonathan.gallim...@gmail.com> wrote:
> Seeing the same with the java11-experimentation build here:
>
> # Created on 2019-03-01T18:01:43.414
> Corrupted STDOUT by directly wri
Seeing the same with the java11-experimentation build here:
# Created on 2019-03-01T18:01:43.414
Corrupted STDOUT by directly writing to native stream in forked JVM 1.
Stream 'java.lang.OutOfMemoryError: Java heap space'.
java.lang.IllegalArgumentException: Stream stdin corrupted. Expected comma
a
Getting this WARNING on :cdi-tomee on buildbot which is taking forever...
# Created on 2019-03-01T20:48:20.010
Corrupted STDOUT by directly writing to native stream in forked JVM 1.
Stream 'java.lang.OutOfMemoryError: Java heap space'.
java.lang.IllegalArgumentException: Stream stdin corrupted. Ex
Buildbot is slow with :cdi-tomee tests causing out of memory warnings. The
PR409 (CXF 3.3.0 and JAXB 2.3.2) updates was the first merge to cause the
slowness. The rolled up merges after show the same.
# Created on 2019-03-01T20:48:20.010
Corrupted STDOUT by directly writing to native stream in
34 matches
Mail list logo