The following is information on how to get SCOM (System Center Operations Manager) 2007 to generate alerts based on the description field of windows event log entries.
1. If you want to filter based on content in the Description field, determine the parameter number that relates to the text you want to find. To find the parameter number, see the following post:
http://blogs.technet.com/stefan_stranger/archive/2008/05/13/opsmgr-2007-parameters-explained.aspx
2. When specifying what to search for, note that the content displayed in the Event Viewer is not always the same as what you need to filter against in SCOM.
E.g. security event id 560, under accesses, event viewer can show WRITE_DAC, but if you filter against parameter 15 for WRITE_DAC it will not work.
The following command will create a text file with all events of type 560, if you then search this file for the specific event you are looking at in event viewer, you can match with what SCOM sees and what the event viewer sees.
LogParser.exe -q:ON “SELECT Strings AS Parameters FROM Security WHERE EventID=560″ > 560.txt
Using the above you can determine that in SCOM that you need to filter for ‘Parameter 15 contains 1539′. You can also use this to find the parameter number (parameter number can be counted from the left starting from 1, not 0).
3. I haven’t had success in filtering based on event type yet (EventType), including it in the filter stopped scom from picking up the alert.
4. Create a test rule in the management pack you are creating the event log monitoring rule(s), which is basic search for application log, Source “User Event” and then use the logevent (http://support.microsoft.com/kb/315410) application to create a test event, to confirm that your management pack is actively running on the target device.
e.g.
logevent -s I -e 45 “Test Event”
Logevent download at: http://www.petri.co.il/download_free_reskit_tools.htm
5. You can also check the ops mgr windows event log to see it has downloaded your changes, but step 4 will tell you for sure. This will also contain information if ops mgr can’t keep up with the event log and is falling behind in monitoring the event log.
6. Monitors are for monitoring for things that affect the health of an object, rules are where monitoring for things that don’t affect object health.
There are 3 types of monitors for the windows event log, manual health reset, automatic after a set time, and automatic after a specific event has been received.
7. Use trial and error, start with basic filtering criteria, check that receive alerts and make it more specific, at each step checking if you receive an alert, being careful to not create a filter that is so generic that it will result in hundreds of alerts.
8. To avoid possible performance issues during step 7, test against a single device by disabling your rule and specifically overriding it to enabled for a specific device, and removing the override and re-enabling once you know it is working if so required.
9. Too many event log monitors could reduce the system wide performance of the monitored device(s).
10. It is possible to filter based on the entire description field, using a parameter with name EventDescription (howto: http://contoso.se/blog/?p=250), but it is recommended to use parameters instead as searching the entire description will have a greater system wide performance impact on the monitored device(s).
[...]