CI error on Windows tests

[zm711]

Error message

RuntimeError: GitHub rate limit exceeded and GITHUB_TOKEN not set; suggest setting GITHUB_TOKEN in order to get increased rate limit

It suggests setting a token. Should we do that to prevent this flaky test failure? @h-mayorquin (you're always the CI king).

See this run for an example:
https://github.com/SpikeInterface/spikeinterface/actions/runs/10719779060/job/29724578046?pr=3370

[alejoe91]

I think that the issue is that at that time there were too many CI jobs in parallel. Let's wait a bit and retrigger

[zm711]

Yeah sono impaziente. Me dispiace. haha.

[h-mayorquin]

Let's keep an eye on this, we have way more runners on neuroconv and this is not a common problem there.

[zm711]

I've only started seeing it in the past month-ish and only on Windows runners. Just so you have what I have seen.

[JoeZiminski]

It might be worth setting a token to remove this as a possibility, the permitted API calls go from 60 per hour unauthenticated to 5000 per hour with a token so should never be an issue once the token is set. I think setting it up is quite simple (famous last words) but needs to be done by the repo / organisation owner.

[zm711]

How about we try to keep a rough count here in this issue and if it gets high enough we convince Alessio ;)

[JoeZiminski]

Yes sounds good!

[zm711]

Have another rate limit on Windows in my PR #4033.

[zm711]

I have seen this happen 3 more times. What do you think @alejoe91 ? Are we close to doing the token?

[alejoe91]

Yep let's do that. There should be a built-in GITHUB_TOKEN so we can just set this in the workflow yaml:

env:
  GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}