not ok 345 - udp_multicast_join (on MacOS)

[frol]

The following tests fails on MacOS:

not ok 345 - udp_multicast_join
# timeout
# Output from process `udp_multicast_join`: (no output)

https://dev.azure.com/conda-forge/feedstock-builds/_build/results?buildId=111682&view=logs&jobId=0c8d6f06-5560-543f-2b4e-92ad5613354a&j=0c8d6f06-5560-543f-2b4e-92ad5613354a&t=ed24c74f-961c-5d1d-8ea1-b9d1b1f99b59

(ref conda-forge/libuv-feedstock#36)

• Version: 1.34.1
• Platform: MacOS

There is an existing issue #2468 with this test failing, but it was automatically closed without a resolution.

[bnoordhuis]

We get reports about that test from time to time but 9 out of 10 times it's the system's firewall or routing rules blocking or dropping multicast traffic.

[frol]

Can the test be improved somehow? It is MacOS build on regular CI; I don't believe there are any specific configurations for the firewall from Conda-Forge side.

[bnoordhuis]

I don't know how but I'm open to suggestions.

[santigimeno]

Can the test be improved somehow?

The tests in their current form are pretty simple. As Ben said, most of the reports ended up being firewall or routing issues.

Anyway, what version of MacOS are the tests run? Can you add some logs around this line to see if it's failing to receive the first datagram or the second?

It is MacOS build on regular CI

Yes, and those test pass: https://ci.nodejs.org/view/libuv/job/libuv-test-commit-osx/

[frol]

I am sorry, but I have little time to maintain the package for Conda-Forge, so I won't be able to dive into it any time soon. Hopefully, it will get resolved eventually. Meanwhile, there is 1.34.0 release there. Thank you for your time!

[bnoordhuis]

Okay, I'll go ahead and close this out. We can reopen if someone has ideas on how to make the test more resilient.