This CL:

https://chromium-review.googlesource.com/#/c/399583/

Depends on these two CLs:

https://chrome-internal-review.googlesource.com/#/c/297397/
https://chrome-internal-review.googlesource.com/#/c/302515/

All three were in the PreCQ as a unit, when the dependent CLs were chumped.

The PreCQ run was canceled (correctly).

However, looking only at the first CL, every PreCQ run appeared to fail with messages like:

The mixed-c-pre-cq trybot for your change timed out after 240 minutes.
This problem can happen if your change causes the builder to hang, or if there is some infrastructure issue. If your change is not at fault you may mark your change as ready again. If this problem occurs multiple times please notify the sheriff and file a bug.

Also, it was treated as a bad PreCQ run, and flags were cleared so that it wouldn't be tested again.

I think that we should have cancelled the run, but not messaged the first CL in a way that looks like an error in our system, and should have retried the first CL without having to re-set the flags manually.