The first Openstack server in Eqiad that was reimaged to Bookworm in T345811 is cloudcontrol1007.
prometheus-openstack-exporter is failing to start:
Oct 31 14:54:01 cloudcontrol1007 systemd[1]: Started prometheus-openstack-exporter.service - prometheus openstack exporter. Oct 31 14:54:01 cloudcontrol1007 sudo[144819]: root : PWD=/ ; USER=prometheus ; COMMAND=/usr/bin/prometheus-openstack-exporter --web.listen-address=:12345 --os-> Oct 31 14:54:01 cloudcontrol1007 sudo[144819]: pam_unix(sudo:session): session opened for user prometheus(uid=106) by (uid=0) Oct 31 14:54:01 cloudcontrol1007 sudo[144819]: pam_unix(sudo:session): session closed for user prometheus Oct 31 14:54:01 cloudcontrol1007 systemd[1]: prometheus-openstack-exporter.service: Main process exited, code=exited, status=1/FAILURE Oct 31 14:54:01 cloudcontrol1007 systemd[1]: prometheus-openstack-exporter.service: Failed with result 'exit-code'.
We didn't have this issue when we upgraded codfw servers to bookworm, because we only run the exporter in the Eqiad cluster, as we don't have a Prometheus instance in codfw (T350010: Deploy 'cloud' Prometheus instance to codfw).
We could still run the exporter in codfw even if there's no Prometheus polling it, just to be able to identify issues such as this one when testing in codfw.