Interesting - that’s on the rumble layer.

so CS only.

why don’t we see that in CI?

Aren’t we running place channel tests or is that flaky?

That’s a flaky test — I haven’t seen that failure before

does drdr always run in an x86_64?

yes (it’s just running directly on the physical machine)

thanks. I will try to have a think about it on this.

I don’t think there’s much to do here; unless you want to try to spend time reproducing the bug

I said that because I have been meaning to understand rumble better and I saw this flaky test as a way into the code. Doesn’t mean I will find the problem, but it’s a way in for me to read this bit of the code at least and understand rumble’s integration into the rest of RacketCS

Here’s a variant that exposes a problem by itself. This one reliably gets stuck on my mahcine, instead of reporting an engine error, but those symptoms could easily be related.