The crate agent gets the list of sensors to query when it’s first initialized, and when slots are deactivated some of the sensors disappear/become unreadable and the crate agent just starts spewing out errors. In order to get it reading out again the agent needs to be brought down and back up so that the smaller subset of sensors is registered. Similarly after hammering/reactivating the slots the crate agent needs to be cycled down/up. The crate agent should be updated so that either: 1) If such errors are caught the sensor list is reacquired automatically or 2) on some fixed cadence the sensor list is reacquired. 1) will fix the issue of losing all sensors when slots are deactivated but will not bring up more sensors when the slots are reactivated/hammered and 2) would fix both issues but could incur some deadtime between the deactivation/activation and when the sensor list is reacquired. @BrianJKoopman do you have any opinions on a good solution for this?