Volume Streaming errors

[m-roskrow]

Hi!

We restart our concourse-workers every morning in order to clear their volumes (we use an initContainer which does an rm -rf on startup). This has not been a problem for a number of months but recently (and without upgrading Concourse in any way) we've started seeing the following errors every morning:
find or create container on worker concourse-worker-2: find or stream volume: failed to stream p2p out from volume: Put "http://<IP>/volumes/d4730afd-93fd-44c3-6d5d-53f06e232fe6/stream-in?path=.": failed to compress source volume: exit status 1
It seems to happen once for random (not all) combinations of volumes / workers, then be ok on rerun.

Its likely something environmental, but I've been looking into it for a while now and can't find anything obvious, so hoping that someone has seen something similar!