How can I know if a Harbormaster unit test failure is a fluke?

I have submitted a review :gear: D154130 which hit a failure on the x64 debian machine: :recycle: x64 debian > MLIR.Examples/standalone::test.toy · B242190 (llvm.org)

I am able to successfully run this test locally on linux, and the CI failure is a timeout. How can I know if this is an infrastructure fluke? Can I ask for a rerun? Should I talk with someone?

You can update your patch and trigger a rebuild, or do it manually in the web interface.
FWIW, the process got killed, could just be a resource problem.

Ah thanks! I didn’t know I had that button. It seems like the rerun passed but now other (non-Harbormaster?) errors showed up and those have a different UI with no rebuild button that I can see :confused: