To Reproduce Steps to reproduce the behavior:
A customer with RHOSP 17.1.8 deployment faced a series of problems coming from libvirt after virtlogd process died for some reason (no clear pointer in logs), while nova_virtlogd container was up and running.

It looks like our nova_virtlogd_wrapper doesn't process such situations properly and doesn't re-create nova_virtlogd container in such circumstances even after tripleo_nova_virtlogd_wrapper is restarted.

Although relevant virtlogd logs are not available, a there is a sosreport collected when virtlogd process was dead, but nova_virtlogd container was running.

Expected behavior
Personally I expect nova_virtlogd container to die after virtlogd process is dead.

Bug impact
Customer spent some time restoring libvirt containers in this situation

Known workaround
Shutdown nova_virtlogd container and restarting tripleo_nova_virtlogd_wrapper should do the needful.

I may have selected incorrect components, please correct me if I am wrong here.