We set up dominant process test to monitor excessive CPU utilization.
The test is set to show "value, process name".
Now we observe that the test sometimes goes to bad with results like:
"109:oracle.exe" or even "1095:oracle.exe"
Any idea?
thomas
Dominant Process test shows CPU load > 100
-
thomasschmeidl
- Posts: 166
- Joined: Sat Apr 15, 2006 2:14 pm
- Location: Germany, Bavaria
-
thomasschmeidl
- Posts: 166
- Joined: Sat Apr 15, 2006 2:14 pm
- Location: Germany, Bavaria
109% is normal value when you have more than 1 CPU or CPU core.
If you have 2 CPU, process can use up to 200% of CPU time (100% on one CPU and 100% on another CPU).
However 1095% value is impossible on 2 CPU system...
We are checking HostMonitor code but do not see any mistake.
Are you sure you setup test to display value? May be it shows process ID?
or may be it shows 195:oracle.exe, not 1095:oracle.exe?
Regards
Alex
If you have 2 CPU, process can use up to 200% of CPU time (100% on one CPU and 100% on another CPU).
However 1095% value is impossible on 2 CPU system...
We are checking HostMonitor code but do not see any mistake.
Are you sure you setup test to display value? May be it shows process ID?
or may be it shows 195:oracle.exe, not 1095:oracle.exe?
Regards
Alex
-
thomasschmeidl
- Posts: 166
- Joined: Sat Apr 15, 2006 2:14 pm
- Location: Germany, Bavaria
We rechecked our code once more and do not see any mistake. Probably target system provides wrong information to HostMonitor.
It can be verified by using WMI Query like
select KernelModeTime from Win32_Process where caption='oracle.exe'
select UserModeTime from Win32_Process where caption='oracle.exe'
If you setup 2 WMI tests with these queries with short time interval (e.g. every 2 seconds) and store all results into log (e.g. private log) then it will be possible to analyze data to find out if system works correctly... however this does not mean we will be able to tell why system works incorrectly.
Is this problem appears on single system?
Regards
Alex
It can be verified by using WMI Query like
select KernelModeTime from Win32_Process where caption='oracle.exe'
select UserModeTime from Win32_Process where caption='oracle.exe'
If you setup 2 WMI tests with these queries with short time interval (e.g. every 2 seconds) and store all results into log (e.g. private log) then it will be possible to analyze data to find out if system works correctly... however this does not mean we will be able to tell why system works incorrectly.
Is this problem appears on single system?
Regards
Alex
-
thomasschmeidl
- Posts: 166
- Joined: Sat Apr 15, 2006 2:14 pm
- Location: Germany, Bavaria
Dear Alex,
we had set up the test on an oracle-db-server, which had inexplainable phases of 100%-CPU. During these phases the server was barely accessible (no rdp possible, slow pings, no response to oracle-clients etc.) and we wanted to find out, whether it was oracle or an other process that caused the problems.
5 days ago we patched the oracle db and have no more 100% CPU phases. I just checked the logs and didn't find any more strange reply in the Dominant Process Test.
Regards
Thomas
you are propably right,Probably target system provides wrong information to HostMonitor.
we had set up the test on an oracle-db-server, which had inexplainable phases of 100%-CPU. During these phases the server was barely accessible (no rdp possible, slow pings, no response to oracle-clients etc.) and we wanted to find out, whether it was oracle or an other process that caused the problems.
5 days ago we patched the oracle db and have no more 100% CPU phases. I just checked the logs and didn't find any more strange reply in the Dominant Process Test.
Regards
Thomas