Breaking this out from #47017 as a narrower issue for tracking and investigation purposes.
There still seems to be some instances where running all.bash sequentially (i.e., without any sharding) on the windows-arm64-10 builder runs into a crash. I've observed one occurrence out of 5 test runs on the go1.17 commit this week:
$ go run . -target=windows-arm64 -version=go1.17beta710 -rev=ec5170397c724a8ae440b2bc529f857c86f0e6b1 -watch
run 1 = "ALL TESTS PASSED"
run 2 = a test known to be flaky (issue #44820) failed: runtime.TestGcLastTime failed: "gc_test.go:126: bad last GC time: got 1629865473098425100, want [1629865473098425100, 1629865473001007600]"
run 3 = "windows-arm64: Error: Buildlet https://farmer.golang.org:443 failed heartbeat after 10.005360314s; marking dead; err=timeout waiting for headers"
run 4 = "ALL TESTS PASSED"
run 5 = "ALL TESTS PASSED"
This problem doesn't reproduce with a physical Windows 10 machine, so this may be a qemu emulation bug. Back when #47017 was filed, this problem happened frequently, but it seems to have become less frequent since then.
CC @golang/release.
Comment From: gopherbot
Change https://golang.org/cl/345129 mentions this issue: cmd/release: start running release tests for windows-arm64
Comment From: dmitshur
We don't have the windows-arm64-10 builder anymore; the current windows-arm64 LUCI builder runs Windows 11, and doesn't involve qemu. This doesn't seem to be happening now, and if it does, it's better to file a new issue. Closing as obsolete.