Stdout from a Swarming shard (https://chromium-swarm.appspot.com/task?id=409ac5e14f925d10&refresh=10&show_raw=1) says:

05:30:00.713 5294 Testing completed. Exit status: 2

Not sure why Swarming thinks it's COMPLETED (SUCCESS)

It's COMPLETED (SUCCESS) because the script is lying in its stdout, the actual exit code is 0, not 2.

I'm not able to reproduce the bug locally. The exit status of r-w-t is consistent with what the log message says locally.

Looking at the commit history of blinkpy, I can see a few changes quickly reverted after landing recently. Perhaps something was broken? We can trigger some new try jobs on ToT and see if we can see the failure mode.

I can still see this pattern this morning, for example: https://luci-milo.appspot.com/p/chromium/builders/luci.chromium.ci/Win7%20Tests%20%28dbg%29%281%29/72381