Starting with cagent version 1.1.0 CPU utilization analysis has been introduced. You might need an update of your agent.
If the load of a system is above a given threshold the most CPU consuming processes are recorded and stored as a snapshot. You can review these snapshots to trace which processes have caused a high load.
CloudRadar stores up to 40 snapshots. If the number is exceeded oldest records are deleted.
CPU Utilisation Analysis is enabled by default and under normal circumstances there is no need to change the configuration. Executing
cagent -p shows the running configuration. You will find the following section.
[cpu_utilisation_analysis]# target value to start the analysisthreshold = 10.0# threshold compare function, possible values: 'lt', 'lte', 'gt', 'gte'function = "lt"metric = "idle"# should be one of values of cpu_utilisation_gathering_modegathering_mode = "avg1"# number of processes to returnreport_processes = 5# how much time analysis will continue to perform after the CPU utilisation returns to the normal valuetrailing_process_analysis_minutes = 5
Processes with the same name are grouped. For example, if the high load is caused by an Apache Webserver you can have hundreds of processes called
apache2. Cagent will group them and sum up their load.
The default means, if the average CPU IDLE of the last minute is less than 10% the CPU Utilisation Analysis is triggered, and snapshots are recorded every time cagent runs. If the threshold is not exceeded any more cpu analysis is never the less active for the next 5 minutes.
All default settings of cagent are not present in the config file
/etc/cagent/cagent.conf (Linux) or
C:\Program Files\cagent\cagent.conf (Windows). If you want to change the settings, you must copy the configuration block as shown above (Lines 1-17) to the configuration file and make your changes. Restart cagent afterwards.