Skip to content
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

Disable "Test spawning a lot of tasks" on Windows #43125

Merged
merged 8 commits into from
Nov 19, 2021

Conversation

tkf
Copy link
Member

@tkf tkf commented Nov 18, 2021

This patch disables the known failure #43124 on Windows for now.

Ref: #43072 (comment)

@DilumAluthge
Copy link
Member

Instead of disabling completely, it might be fine to just dump all the output to the log. Then, the next time we see this failure in CI, we can actually collect some information on why it failed.

@DilumAluthge
Copy link
Member

I also don't object to disabling it completely on Windows. Either approach is fine.

@tkf
Copy link
Member Author

tkf commented Nov 18, 2021

Yeah, maybe we can just pull only the first commit. I guess it'd be easier to notice the failure if the CI actually fails.

@tkf
Copy link
Member Author

tkf commented Nov 18, 2021

Let's pull in the first commit via #43126 first. Then I'll prepare a patch to disable the Windows test (without the logging message #43125 (comment)) in this PR. This way, we can disable the Windows test with a few clicks.

@tkf tkf marked this pull request as draft November 18, 2021 08:03
@tkf
Copy link
Member Author

tkf commented Nov 18, 2021

Never mind, this is pretty reproducible. I get the log output

┌ Error: Failed test: "spawn and wait *a lot* of tasks in @profile" (n = 2000000)
│ 
...
│   proc.exitcode = 3221225477
│   proc.termsignal = 0

in https://build.julialang.org/#/builders/65/builds/5543

3221225477 is STATUS_ACCESS_VIOLATION so there's something wrong. But it looks like there's no backtrace.

Note that n = 2000000 is the largest number of tasks in the test. I wonder if ignoring this case is sufficient.

BTW, #32938 would have been helpful for noticing this from the vanilla test output.

@DilumAluthge
Copy link
Member

DilumAluthge commented Nov 19, 2021

Note that n = 2000000 is the largest number of tasks in the test. I wonder if ignoring this case is sufficient.

Let's go ahead and give this a try?

@tkf tkf marked this pull request as ready for review November 19, 2021 01:28
@DilumAluthge DilumAluthge added the merge me PR is reviewed. Merge when all tests are passing label Nov 19, 2021
@tkf
Copy link
Member Author

tkf commented Nov 19, 2021

So, there seem to be a failure with n < 2000000: https://build.julialang.org/#/builders/65/builds/5597/steps/5/logs/stdio

Let's skip the tests on Windows. Decreasing n maybe could make it less frequent but it'd make it just more annoying.

@DilumAluthge DilumAluthge added merge me PR is reviewed. Merge when all tests are passing and removed merge me PR is reviewed. Merge when all tests are passing labels Nov 19, 2021
@vtjnash vtjnash merged commit 672c641 into JuliaLang:master Nov 19, 2021
@tkf tkf deleted the skip-windows-spawnmany branch November 20, 2021 00:18
@DilumAluthge DilumAluthge removed the merge me PR is reviewed. Merge when all tests are passing label Nov 20, 2021
LilithHafner pushed a commit to LilithHafner/julia that referenced this pull request Feb 22, 2022
* Don't hide output from spawnmany test script
* Disable "Test spawning a lot of tasks" on Windows
* If `!success(proc) || timeout`, print a variety of debugging data

Co-authored-by: Dilum Aluthge <dilum@aluthge.com>
LilithHafner pushed a commit to LilithHafner/julia that referenced this pull request Mar 8, 2022
* Don't hide output from spawnmany test script
* Disable "Test spawning a lot of tasks" on Windows
* If `!success(proc) || timeout`, print a variety of debugging data

Co-authored-by: Dilum Aluthge <dilum@aluthge.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants