When a “cmk -R” does not help anymore, you can try to stop the whole site (omd stop), kill all hanging processes (or even reboot the host), and start the site again.
If a site is really messed up and cannot be repaired, for whatever reason: a remote (slave) site in a Distributed WATO setup is essentially dumb. You can always shut it down (omd rm), create a new site (omd create) and reconnect the central site to it. You will just lose the historic monitoring data.