Monitoring system cpu utilization

3 minute read

Posted 2021-06-26 - Updated 2021-07-04
Supported on Linux, Aix, MacOS


On this page we will demonstrate how you can monitor the system cpu utilization with the System Monitor. We demonstrate with a practical example how to do it.


Monitoring the CPU utilization

  • You can be alerted if the usage percentage of your CPU reach a certain threshold for a certain number
    of minutes, specified in column 6.
  • Every time a “cpu_level” line is process by the System Monitor, it used the “vmstat” (iostat on MacOS) command to get a snapshot of the cpu utilization (user + system). The resultant value is then compare with the warning and error threshold specified respectively in column 4 and 5.
  • If the value is under the warning and error level, then the event starting date (‘YYYYMMDD’) and time (‘HHMM’) (Column H’ and ‘I’) is set to zeroes (‘00000000 0000’).
  • If the value is over or equal, depending of the test you have put in column 3, then two things can happen.
    • If the event starting date and time (Column H’ and ‘I’) are all zeroes, then this is the first time the cpu usage exceed the warning or error level. Then the System Monitor set the event starting date and time to the current date and time of the system and processing continue to the next line.
    • If the event starting date and time (Column H’ and ‘I’) are not all zeroes, then this is not the first time the cpu usage exceed the warning or error level. Then the System Monitor calculate the number of minutes since the event start date and time.
      • If the number of minutes is less the number of minutes specify on column 6, then processing continue with the next line.
      • If the number of minutes is greater or equal (test in column 3) than the warning or error threshold value, then the warning (Column J) or error group (Column K) will get alerted depending on which value is exceeded and then the event starting date and time (Column H’ and ‘I’) is set back to zeroes (‘00000000 0000’).
# ID COLUMN 1  2  3    4   5  6   7    8   9 A B C D E F G     H     I     J      K    L
cpu_level      80 >=  85  95 120 0700 2100 Y Y Y Y Y Y Y Y 20210601 1445 wargrp errgrp -

In the example above, the current cpu usage is 80%, the warning threshold is set to 85% and error at 95%. If the percentage of utilization is greater or equal than one of these value for more than “120” minutes (column 6), then the warning group “wargrp” or the “errgrp” will get alerted.


Example of SysMon output In this example, the CPU usage have reach 100%. Sysmon have calculated that it has been over the error level (95%) for 36 seconds, so if it reach 1800 seconds (30 minutes) an alert will be automatically generated.

# sudo sadm_sysmon.pl
Checking CPU Usage ...
CPU Usage line:   6  0 211712 311712 588488 1526080    0    0     1   134 6373 13407  5 95  1  0  0

CPU User:   5 - System:  95  - Total: 100
 - Warning Level: 85 - Error Level: 95
Actual Time is 2021 07 05 09 00 36
Actual epoch time is 1625490036
Load on cpu started at 2021 07 05 09 00 00 - 1625490000
So 1625490036 - 1625490000 = 36 seconds
You asked to wait 1800 seconds before report an error

...
...

Back to the top

See also

Link to … Description
sadm_sysmon_tui.pl Command line summary of alerts and failed scripts of all your servers.
sadm_sysmon.pl Client system monitor
sadm_fetch_clients.sh rsync all .rch/.log/.rpt from actives clients to the SADMIN server
SysMon configuration file Client System Monitor configuration file
sadmin.cfg SADMIN main configuration file

You may also enjoy