Under normal conditions, monitor queue size to establish a baseline that can be used to detect performance issues early.
To set up automatic monitoring of queues, a script may be written to parse the output of the dump queues. The script could specifically parse the current size for each queue. For example, if the current size exceeds a specified threshold, then alert someone to investigate this further. Some queues would need to be exempt from the threshold, such as ICS_ActionScheduler.
A possible threshold for queues would be 10,000. However, this setting would depend on your specific needs. It would also be possible to set the threshold lower. If you set a lower threshold, you should do more monitoring to determine whether or not there is a consistent queue growth.
If queue growth happens consistently over time, further investigation is recommended.