Note: This is the first betty-arcnext-chrome-pfq that started timing out:

https://cros-goldeneye.corp.google.com/chromeos/healthmonitoring/buildDetails?id=2951782

Also, betty-arc64-release appears to be timing out in VMTest as well:

https://cros-goldeneye.corp.google.com/chromeos/legoland/builderHistory?buildConfig=betty-arc64-release&buildBranch=master

https://chromium-review.googlesource.com/c/chromiumos/third_party/autotest/+/1226161 just exempts a script that is running anyway from a security test, I think it is unlikely to be the culprit.

It looks like there is a crash somewhere confusing tradefed, even though it looks like everything [151/151] is passing.


09/18 03:21:34.387 INFO |             utils:0287| [stdout] 09-18 10:21:34 I/ConsoleReporter: [151/151 armeabi-v7a CtsAccountManagerTestCases 127.0.0.1:9227] android.accounts.cts.AccountManagerTest#testNewChooseAccountIntentDepracated pass
09/18 03:21:34.387 INFO |             utils:0287| [stdout] 09-18 10:21:34 I/ConsoleReporter: [127.0.0.1:9227] Test run failed to complete. Expected 151 tests, received 20. onError: commandError=false message=INSTRUMENTATION_ABORTED: System has crashed.

I guess #5 is b/115944638?
This type of failure was not expected to be surfaced as failure, but b/115944638? turned it to FAIL (and I already landed a remedy.)

Yes, but apart from that, the crash is concerning.
I at least found 3 different crashes hidden behind and started analyzing them on b/115944638. Perhaps more to come.

Teardown-time crash is observed elsewhere b/116009991#comment9
and for that one the fix has landed b/115949068 (which is one of the bugs split out from b/115944638).

#5 may be the same one.
(I'm always having hard time to find the actual failing test from VMTest, so I haven't looked into the failure log yet, though...)

Yeah
https://cros-goldeneye.corp.google.com/chromeos/healthmonitoring/buildDetails?id=2951782
https://cros-goldeneye.corp.google.com/chromeos/healthmonitoring/buildDetails?buildbucketId=8935069237701676848
those were due to b/115949068.


Summary for #comment5:

https://chromium-review.googlesource.com/c/1229814/   (R71-11078.0.0)
should be able to mask the crash as warning, rather than a failure.

Fix for the root cause (the system crash) is fixed on ToT ARC-P image.
We though need a green Android PFQ for the change to be propagate to the Chrome OS tree:
https://bugs.chromium.org/p/chromium/issues/detail?id=884828


tradefed looks recovered from the crash and run [151/151] cases but one of them looks to have crashed. (I/ConsoleReporter are somewhat sorted and confusing.)

Not sure about the original timeout issue, but if that crash+retry was the problem, green Android-pi PFQ should fix the trouble, I hope.

Don't worry, if it doesn't fix the issue we can just swap this out for another stable test. But for consistency it might be worth waiting for the fixes and keeping it as it is.