Help debugging cancelled runners on main branch

I’ve started to notice some runs marked as “Failed/Failing” which aren’t.

One job has 6 parallel runners and for some reason some runners are getting cancelled. This is on the main branch so the “auto-cancel” feature should not apply. I don’t see any timeouts being hit- the runs finish in around 30mins and get consistent output. Nothing has changed that I know of but this started around July 7th.

So this could be 2 separate problems:

  • Parallel runners are getting cancelled on the main branch
  • Cancelled runners should are showing as “failed”