Finally worked around github’s problem with environment variables, but not something fails and github provices a less than helpful message: https://github.com/racket/racket/runs/615277002?check_suite_focus=true

lol

I merged anyway because the problem we were having is fixed.

Still, going to start annoying the hell out of the github engineers.

Come on… like this one : https://github.com/racket/racket/runs/615511739#step:14:131

Now it says the operation was cancelled. Who cancelled? GitHub?

It’s like they are running these things on some sort of Amazon EC2 Spot instances and if the instances are taken down by amazon, they just cancel the run. A bit ridiculous.

Of course, I don’t know where the jobs are running but that’s the feeling I get.

it definitely runs on Azure, but I assume they have something like Spot

Right, I forgot this is all Microsoft domain… so azure.

This happens again: https://github.com/racket/racket/runs/617049509

From what I understand, it means there’s an internal error.

But they really should provide some information…

I don’t know if this is possible, but can we give rights to PR-ers to restart CI?

For example, my PR (https://github.com/racket/racket/pull/2700) is failing because I pushed the commit when GitHub services were failing a couple of days ago. I want to run the tests again, but I don’t have a permission I think.