Bug #16920: TestThread#test_signal_at_join fails on aarch64

• Status: Closed
• Assignee: -
• Target version: -
• Backport: 2.5: UNKNOWN, 2.6: UNKNOWN, 2.7: UNKNOWN

[jaruga (Jun Aruga)]

I observed the following error on Ruby 2.7.1 building in Fedora aarch64 CI. It was just one time error. It does not always happen.

  1) Error:
TestThread#test_signal_at_join:
Timeout::Error: execution of assert_separately expired timeout (120 sec)
pid 2314088 killed by SIGABRT (signal 6) (core dumped)
| 
    /builddir/build/BUILD/ruby-2.7.1/test/ruby/test_thread.rb:1346:in `test_signal_at_join'
Finished tests in 1379.712832s, 15.2372 tests/s, 1972.9127 assertions/s.

[mame (Yusuke Endoh)]

I have never seen the test fail on Debian/Ubuntu aarch64. So use Ubuntu! (joke)

It is difficult to investigate the issue with no access to the machine...

Can you reproduce the issue with ruby master? Can you try another Fedora+aarch64 environment? Can you measure how long the following command takes on the machine? (It takes less than one second on my machine, so I guess extending timeout would not solve the issue.)

make test-all TESTS="test/ruby/test_thread.rb -v -n test_signal_at_join"

[jaruga (Jun Aruga)]

I have never seen the test fail on Debian/Ubuntu aarch64. So use Ubuntu! (joke)

I love Debian. :) It's my pleasure when I spend time with people in Debian community.

It is difficult to investigate the issue with no access to the machine...

Yeah, it makes sense.

Can you reproduce the issue with ruby master? Can you try another Fedora+aarch64 environment? Can you measure how long the following command takes on the machine? (It takes less than one second on my machine, so I guess extending timeout would not solve the issue.)

Yes, I will try to reproduce the issue on the better way to inform you later, after my current focusing tasks.

[vo.x (Vit Ondruch)]

Still issue with ruby 3.1.0dev (2021-12-07 master ec878dac90) [aarch64-linux]

[jeremyevans0 (Jeremy Evans)]

• Status changed from Open to Closed

I reviewed every recent RubyCI arm/aarch64 failure, and this test did not fail in any of them. So failures must be rare. I'm going to close this. If this is still an issue and someone can come up with a reproducible test case, that would be appreciated.