adoptium / adoptium/aqa-tests

AIX/HS/JDK11 MiniMix failing on weakCompareAndSet() operation on POWER9

Open
#2,332 17 comments 0 reactions 0 assignees View on GitHub
aix JBS
Dominant language
HTML
Stars
155
Forks
346
Avg merge
2d 9h
Merged PRs (30d)
36

Description

**Describe the bug**
On the `MiniMix_5m_1` test in the `system` category (although it has also been seen on [ConcurrentLoadTest_5m_1](https://ci.adoptopenjdk.net/job/Grinder/7373/))we are getting this failure periodically. I'm doing various tests to figure out when it's occurring but so far I've only seen it on the only POWER9 system that we have ([test-osuosl-aix71-ppc64-1](https://ci.adoptopenjdk.net/computer/test-osuosl-aix71-ppc64-1/))
```
LT >>> Captured test output >>>
LT testStarted : testAPI(net.adoptopenjdk.test.concurrent.atomic.AtomicLongArrayTest)
LT testFailure: testAPI(net.adoptopenjdk.test.concurrent.atomic.AtomicLongArrayTest): 27 : weakCompareAndSet() expected: but was:
LT junit.framework.AssertionFailedError: 27 : weakCompareAndSet() expected: but was:
```

**To Reproduce**
Steps to reproduce the behavior (or Grinder rerun link): https://ci.adoptopenjdk.net/job/Grinder/7395

**Expected behavior**
Test pass :-)

**Additional context**
Add any other context about the problem here.
I've queued up various grinders after the initial tests I did yesterday in order to test various options to determine the extent of it. So far I've only been testing on JDK11 HotSpot - mostly these are five iteration runs. I'll add notes as they complete.
- JDK11/HS again: https://ci.adoptopenjdk.net/job/Grinder/7394 (3/5 iterations failed)
- JDK11/HS yet again :-) https://ci.adoptopenjdk.net/job/Grinder/7395 (5/5 iterations failed)
- JDK11/J9: https://ci.adoptopenjdk.net/job/Grinder/7396 (0/5 iterations failed)
- JDK16/HS: https://ci.adoptopenjdk.net/job/Grinder/7397 (2/5 iterations failed)
- JDK11/HS with `-Xint`: https://ci.adoptopenjdk.net/job/Grinder/7398 (Passed)
- JDK11/HS with `timeLimit=1m` instead of `5m`: https://ci.adoptopenjdk.net/job/Grinder/7399 (Passed)

I've also queued up some more 10 iteration runs on other POWER8 machines just to get some extra validation as to whether it is specific to the POWER9 system:
- [test-osuosl-aix72-ppc64-2](https://ci.adoptopenjdk.net/job/Grinder/7400/console) - 0/10 iterations failed
- [build-osuosl-aix71-ppc64-2](https://ci.adoptopenjdk.net/job/Grinder/7401/console) - 0/10 iterations failed
- [test-ibm-aix71-ppc64-2](https://ci.adoptopenjdk.net/job/Grinder/7402/console) - 0/10 iterations failed
- And https://ci.adoptopenjdk.net/job/Grinder/7403/console on which ever machine it happens to land on :-) - 0/10 iterations failed

FYI @lumpfish as we were discussing this

Contributor guide

Open the contributing guide

Research direction

Start with the AtomicLongArrayTest testAPI failure in the MiniMix_5m_1 run and reproduce it through Grinder 7395. Compare the listed JDK11 HotSpot, JDK9, -Xint, time-limit, POWER8, and POWER9 results; done means the weakCompareAndSet assertion passes reliably on the affected configuration.

Written by the indexing model from the issue text.

Assessment

Tech stack
java
Domain
operating-systems, testing-qa
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Mostly clear
Newbie friendliness
35/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.