To Reproduce Steps to reproduce the behavior:

1. Using install_yamls start ironic either standalone or as part of an openstack deploy
2. Run the associated cleanup to stop all services
3. ironic-conductor-0 continues to run until the timout and it is given the SIGKILL signal

Expected behavior

• ironic-conductor-0 pod will stop instantly with a graceful shutdown on SIGTERM

Bug impact

• Any stop or restart of the conductor service will be very slow due to the service not stopping in a timely manner

Known workaround

• Waiting for the timeout

Additional context

• I'm not sure if my graceful shutdown rework is in Antelope. If it is then maybe there is a regression. If it isn't then some Antelope only fix may be required.