This message was deleted.

update: jacking up to three nodes seem to solve this. If it’s a resource problem though, I’m curious why waiting 5 seconds in between still works

Having an egress instance hit 100% CPU will cause every recording to fail, so they will not accept multiple requests at the same time