niscache folder on robot was 6 GB and huge q1.rdb/q2.rdb files
-rw------- 1 root root 3.9G Feb 28 09:30 q1.rdb
-rw------- 1 root root 3.0G Feb 28 09:29 q2.rdb
The robot was using the "oracle" probe which was monitoring many hosts.
The oracle databases had failed over, which resulted in many alarms and QOS.
Concurrently, the robot had lost its connection with the hub.
UIM 8.51 MS SQL 2012
Adjust the "Message Queue Size Limit" or other parameters on the spooler on the robot.
Message queue size limit (KB) - specify the maximum size of the message queue in KB. Every hour, the spooler checks the size of the message queue. If the size limit is exceeded, the spooler shrinks the queue to the specified limit. The oldest messages in the queue are removed first.