I’m getting the above same error message on my Kubernetes cluster with my personal Discourse set up on the latest main branch of the plugin and the (as of this post) latest 2.8.0.beta4 version tag. I can provide steps to reproduce this locally if anyone is interested.
If I use the stable tag or the stable-branch I get something similar to the following sequence:
Attempting to kill pid 927
Prometheus Exporter, failed to send message Connection refused - connect(2) for "localhost" port 9405
Detected dead worker 927, restarting...