hey, just ran into this issue: 1. I run pants test :: and it kicks off a bunch of parallel tests via subprocesses. 2. One of the tests fails during requirements.pex building due to a missing dep (like we run our own cheeseshop and don’t fall through to pypi - it’s not great, but we have reasons for this) 3. Pants exits 4. The subprocesses stay active as orphaned procs Eventually they die. Now I don’t know (since i lose their stdout/err) if they die because they also ran into issues, or they completed, or if pants is just delayed in killing them. Since they’re owned by init I feel like it’s the former, in which case pants isn’t cleaning them up properly. Just thought I’d flag. Seems like a kill to the pgid would solve this after handling the initial subprocess exception. Sorry, I don’t have a great test case for this.

Hey Nate, thanks for flagging. What pants version is this?

2.2.0

cc @witty-crayon-22786

thanks for the report!

yah the ppid was 1 for them

is

pantsd

enabled…?

yah

interesting! do you still have the

pantsd

log from that run?

.pants.d/pantsd/pantsd.log

?

let me try to clear it and redo it since it’s quite full from other stuff

thank you!

gtk thx

…i suspect that i know what happened, but the log will confirm!