CI: Only run on pushes to main & cancel stale executions

[saschanaz]

So the SIGABRT thing is some Rust compiler regression?

[badboy]

So the SIGABRT thing is some Rust compiler regression?

oh here as well? It might. I think it might be running into memory limitations maybe, at which point github actions probably kills it? That's just an assumption, I don't actually know if that's the cause.

[saschanaz]

Hmm, are macOS jobs getting less memory? I thought it was constant at 2GB.

[badboy]

Hmm, are macOS jobs getting less memory? I thought it was constant at 2GB.

Maybe? Probably? Memory usage still differs between operating systems. Over on Glean we had issues with memory usage spikes caused by different Rust versions in the past.

[saschanaz]

For now I retriggered the failing job. I also wonder we have any stress-testing test here, in that case perhaps it's better to split it somehow to a separate job. Very weird to hit the resource limit if we don't have one.

[saschanaz]

Some Node.js test frameworks like Jest has an option to show the memory usage on each test, does Rust have an equivalent? 🤔

[badboy]

For now I retriggered the failing job. I also wonder we have any stress-testing test here, in that case perhaps it's better to split it somehow to a separate job. Very weird to hit the resource limit if we don't have one.

Again: that was just my suspicion. I can't say for certain that this is the cause.

[saschanaz]

Retriggering made it green! (which is... good and bad, I hate intermittent failures 🥲)