Cheshire Cat Computing

A reported problem, thought to be fixed in v1.6.2 but apparently not, exists where under some conditions, NagEventLog will start to take up all available CPU time. This may happen a day or more after the service is started, however so far we have not been able to duplicate this problem.

It may only occur under windows 2003. V1.6.2 works under 2003 (it uses the winsock2 stack instead of the winsock1 stack used by 1.6.1 and earlier)

Two users have reported this problem so far.

If anyone else is experiencing this, please post here and specify
1) which version of NagEventLog you are using (latest is 1.6.2)
2) which version of Windows you are using it under
3) if you can reproduce this at will - eg, it happens immediately, or on the first test message logged
4) If you can successfully send two successive test NSCA messages from the configuration tool.

Thankyou for your help!

_________________
Steve Shipway
UNIX Systems, ITSS, University of Auckland, NZ
Woe unto them that rise up early in the morning... -- Isaiah 5:11

Hi,
i had Nageventlog 1.9.2 in use and i had a CPU consumption problem.
Every time NagEventlog is running it's check the cpu is consumption was very high.
At the end, the CPU Load was high during the complete time and it never goes down!
I had many W2K3 Domain Controllers with SP2 in use and on the most of them
i had this Problem.
On many other normal W2K3 Server i'd never seen this Problem.

So, i'd investigate this Problem.
The first Step what i done was to increase the Check period.
This was a little solution for the Problem, because when the check starts and uses manny CPU
it takes a long time. And normally the time goes by and the next check should be startet,
but the first check was originally running with high CPU load.
So, now with the extened check cycle time, the first check will be finished and i see that now the
cpu load goes down until the next check cycle starts.

The next what i found was that on normal W2K3 Server the Windows Eventlog Buffer size is normally round about
16MB. ( e.g. system eventlog or security eventlog )

Now i checked on Active Diretory Domain Controllers the size of the Eventlog Buffers.
I found that the Windows Security EventLog Buffer in my case was 128MB !!!
I decreased the size of the security Eventlog Buffer to 16MB and erased it,
so after that the new size is in use.

Now i restarted the NagEventlog and checked the CPU consumption.
The CPU-Load was now nearly nothing for this processs !!!
After the Security Eventlog was filled with entries the CPU load for this process is not raising up !!!

So, i hope this will help other people which had the same Problem...