When user is using request.cfg - if one of the probe listed does not exist the robot stops responding until the problem is resolved and the "NimbusWatcherService" service is restarted.
Looking for a way around this - as, failing in this manner instead of just sending an alarms is not very productive.
Release : 20.3
Component : UNIFIED INFRASTRUCTURE MGMT
Using request.cfg for probe deployment (following the KB34728) is the older way of doing it, which might not be tested completely with UIM 20.3.
One option will be to use MCS Group Profiles to deploy probes to a list of Robots on UIM 20.3.
Another option will be configure Robots based on OS. This way, you can setup MCS Profile, and then - if the robot is Windows deploy x probes, if Linux deploy y probes.