> ideally we would know why it takes 40s+ for ...
# corem
Mike Myers
10/21/2024, 10:04 PMideally we would know why it takes 40s+ for the internal HTTP server to start serving in Sequoia.Did we try using
127.0.0.1localhosts
Stefano Bonicatti
10/21/2024, 10:35 PMAh I see, did not test
Stefano Bonicatti
10/21/2024, 10:38 PMBut I launched a build with that change on my repo, so will see if it makes a difference, good point!
🙌 1
Stefano Bonicatti
10/21/2024, 11:14 PMUnfortunately it did not have an effect
m
Mike Myers
10/21/2024, 11:17 PMAh! So this issue is observed on the hosted macOS 15 runners only though? Next thing would be to know if it happens on a local dev VM
s
Stefano Bonicatti
10/21/2024, 11:22 PMThat's where I can test macOS 15 for now. I haven't updated yet my machines; I have installed macOS 15 on a VM but still did not manage to fully configure it to do the test
m
Mike Myers
10/22/2024, 12:05 AMIf I’m understanding the CI logs, it’s this line in the Python that takes like 30 seconds to complete: https://github.com/osquery/osquery/blob/cd006f65b9ec3aa3022913a02826bf31e15c349b/tools/tests/test_http_server.py#L394C5-L394C68 and that’s just instantiating the HTTPServer object
Mike Myers
10/22/2024, 12:09 AMOr rather, it takes that long time to reach the following line, after launching the Python process. I wonder also if it could be a Rosetta delay if it’s launching the x64 Python interpreter for the first time and there’s some kind of first-time Rosetta initialization going on?
s
Stefano Bonicatti
10/22/2024, 1:09 AMI originally tried to update the python version and fix the architecture (the PR you reviewed), but that wasn't it either
🤔 1
Stefano Bonicatti
10/22/2024, 2:50 PMI just tried in a VM, but I don't have the issue, but the CI is still using 15.0, not 15.0.1 and I seem to recall that it had some network issues indeed
m
Mike Myers
10/22/2024, 4:59 PMoh, and Apple is no longer distributing 15.0 I bet
6 Views