Hypnotoad unexpectedly stopped after killing a busy worker

[aschweig]

I am running Hypnotoad via systemd; I had an issue during a zero-downtime upgrade where one worker was blocking and I eventually saw "Stopping worker 28125 immediately" in the log. Two seconds later, I see from journalctl that the systemd service logged SERVICE_START followed by SERVICE_STOP -- and at that point hypnotoad and the service were stopped. Is it reasonable to suspect that stopping a worker immediately would cause the service to stop? If so, can I change that? Should I be looking elsewhere?