Currently the on-prem networking services only have logs from the currently running instance. The previous log file is always empty. I assume this is because they are static pods so the logs are just pulled from the running container, but this can be problematic for debugging certain types of problems because we may need logs from before a reboot or restart.

The suggestion that came out of a discussion of one such problem was to redirect the console output to /var/log and then add a gather-extra script that collects those files.