-
-
Notifications
You must be signed in to change notification settings - Fork 310
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
AIX: Grinder target cmdLineTester_jvmtitests_5 fails because dependent shared libraries are not found #2264
Comments
Thanks for the report @aixtools, I will look at how the cmdLineTester_jvmtitests_5 test is configured to see what our options are for changing test config. |
Rerunning cmdLineTester_jvmtitests_5 in a Grinder to get a fresh set of results: |
Interesting, based on Grinders above, this test passed in the jdk11 run on test-osuosl-aix72-ppc64-1, but failed in the jdk8 run on same machine. |
Yes - very interesting. I'll try as well - with save workspace - and run my traces/trcrpt to see what OS behavioral changes I can find. |
Placeholders: https://ci.adoptopenjdk.net/job/Grinder/7290/ (jdk8) and https://ci.adoptopenjdk.net/job/Grinder/7291/ (jdk11) Marked with 'Keep Build' so the logs don't disappear so quickly. Note: remember to release when finished. Forever, is a very long time :) |
I looked at your console outputs - at the surface level the commands seem the same - so what does changing jdk version change in terms of what actually get executed? Helps to know what/where to look. |
Okay it actually looks like there is no native testimage package for jdk8 being found and downloaded via API (and therefore the message, jdk11 case:
jdk8 case:
|
We need to look and see why the native test image for jdk8 AIX is not found, is it being built? From the build job (example: Build 980), looks like they are built. Next thing to check, are they being moved correctly to the releases repo? |
Well, this seems to explain why the test fails. A giant step forward. |
Looks like originally testimage is not available for jdk8 https://github.com/AdoptOpenJDK/openjdk-tests/blob/master/get.sh#L225 #1661. Need to update to support testimage for jdk8. |
#2320 is now merged (and fixes the originally reported issue). Verifying via this Grinder: https://ci.adoptopenjdk.net/job/Grinder/7422 It may be that there is a new different issue for this test target, but that should get raised separately. |
We now have the the native test images, but we see that the test causes a crash based on output from Grinder/7422, not investigated why or if it has been seen at ci.eclipse.org/openj9 (@pshipton have you seen this before?):
|
I don't see any matches. |
There is an internal workitem (RTC 145045: [aix_ppc-64_cmprssptrs] SE80_GIT J9vmTest_2 crash -Xlp16G Requested page size in GC logs matches with page size in -verbose:sizes output) , which is a regression being investigated but probably introduced a while ago just occurred more recently.
Note: the internal test failure is at |
Actually I think they might be related
This failure is |
The problem Jason refers to (145045) was introduced Feb 17th by eclipse-openj9/openj9#11305 and/or eclipse/omr#5652. It would be interesting to see the result of
|
145045 also only occurs when @zl-wang fyi |
Describe the bug
This test fails to load - and grinder reports unstable as result.
The java executable (or .jar code) defeat the normal AIX rtld behavior.
Note: JAVA messages - with an AIX rtld message embedded - see below for a pure AIX example of rtld messaging
To Reproduce
Steps to reproduce the behavior (or Grinder rerun link):
The key command - from a Grinder job - with workspace saved:
"/home/jenkins/workspace/Grinder/openjdkbinary/j2sdk-image/bin/java" -Xgcpolicy:metronome -Xcompressedrefs -Xdump -agentlib:jvmtitest=test:rnwr001 -cp "/home/jenkins/workspace/Grinder/openjdk-tests/TKG/../../jvmtest/functional/cmdLineTests/jvmtitests/jvmtitest.jar" com.ibm.jvmti.tests.util.TestRunner
Expected behavior
That the test runs and finishes with status 0
Screenshots
If applicable, add screenshots to help explain your problem.
Additional context
To assist with understanding the error message above:
JVMJ9TI001E Agent library jvmtitest could not be opened
: This bit is from java - what follows (need slave machines to set up build #2 and 3) is the AIX rtld messages - note: jvmitest is a .jar fileSystem error: No such file or directory)
: the error at OS levelJVMJ9VM015W Initialization error for library j9jvmti29(-3)
: (java message)To further help reading: an example using a simple application (/usr/bin/python)
PYTHON has the following library dependency:
The first library it needs is
/opt/freeware/lib/libpython2.7.so
. It's dependencies are:To get a similar message as this test - rather than rename libpython2.7.so I rename one of it's dependencies - and then try to start python.
Note: this message is known to IBM - but for a differnt case:
My first thoughts is that it can be resolved by xml files - currently not available in the test directories:
Or, perhaps an additional argument - something like -Xbootclasspath/p:/home/jenkins/workspace/Grinder/openjdkbinary/j2sdk-image/jre/lib/ppc64/compressedrefs/ (I do not know java - so this may include a syntax error)
The text was updated successfully, but these errors were encountered: