-
Notifications
You must be signed in to change notification settings - Fork 2.7k
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
io.quarkus.smallrye.graphql.deployment.ConcurrentAuthTest fails on MacOS M1 #28033
Comments
I've seen this error happen sporadically with other tests as well. Can you reproduce this issue when running this single test only? |
Yeah I wonder if we are hitting the infamous file descriptor exhaustion issue you all had with macOS. |
I didn't raise a defect until I'd reproduced locally .... but now I can't reproduce locally (even after un-disabling). So I suspect Guillaume may be right about the file descriptor issue. |
I just built the graphql extension on my M1 - no problem. |
I un-disabled the test, ran a build, and same thing - the build overall was a bit unhappy and hung. That complicates diagnosis, but https://github.com/quarkusio/quarkus/actions/runs/3076919532/jobs/4971564330 shows
(When I had the test disabled, I had a CI run which was clean apart from #28035. So there's clearly some intermittency, and possibly a machine-specific issue, but it seems to be specific to that test (based on the sample of 4).) |
Also specify Mac OS Reverse disabling test Re-disable test
the It would be interesting to run the build with a profile or collecting JFR events (there is probably an even tracking open / close files) You may need to configure your |
Also specify Mac OS Reverse disabling test Re-disable test
I agree that it happens at that point because we hit the limit, but the interesting thing is that we don't hit the limit at all without that test. So that test/underlying code must be particularly hungry (or leaky) in its file consumption. The good news is it perhaps narrows down the scope of what we will need to profile, following @cescoffier's suggestions. This issue can track the investigation to decide whether there's something we can/should change on our side, or whether the test and implementation are hungry-but-correct and we need to adjust the machine configuration. |
That is weird, I would need to track how many file descriptors are opened by this test or if we have a leak. |
Also specify Mac OS Reverse disabling test Re-disable test
Also specify Mac OS Reverse disabling test Re-disable test
Also specify Mac OS Reverse disabling test Re-disable test
Also specify Mac OS Reverse disabling test Re-disable test
Also specify Mac OS Reverse disabling test Re-disable test
Also specify Mac OS Reverse disabling test Re-disable test
Describe the bug
Note: This test has been disabled on M1 as part of #27156
This is the error:
Expected behavior
No response
Actual behavior
No response
How to Reproduce?
Undo the disabling of the test, and then
Also can be reproduced on the Mac CI.
Output of
uname -a
orver
No response
Output of
java -version
No response
GraalVM version (if different from Java)
No response
Quarkus version or git rev
No response
Build tool (ie. output of
mvnw --version
orgradlew --version
)No response
Additional information
No response
The text was updated successfully, but these errors were encountered: