Well, topas -C is talking to the xmtopas data provider daemon on each LPAR its monitoring.
If there are zero CPU cycles then the daemon is struggling to get CPU time and failing to respond regularly to topas that you are watching.
There is not much topas can do about this.
The same problem would happen with any performance monitoring tool.
You could try boosting the priority of the daemons by changing their nice value but as they are not running most of the time, it might not help much - i.e. they should already be fairly high priority.
nmon only monitors a single AIX or Linux - so it will not give you a Global machine view
- although there is a option to boost its priority for exactly this reason (if you are root) see the -Z option.
Ganglia would be lighter weight than topas as it collects data roughly once every 30 seconds.
But if you have zero CPU cycles spare anything is going to struggle to monitr your sydtem.
Hope this helps, report back please if you find a slick solution, ta Nigel
