Hi,

I'm looking to use Graylog 1.3 to find and display potential error states 
across our Windows server estate.

The technique I had hoped to use is "throw out all logs that are evidence 
of normal behaviour; everything that remains must be interesting".  

Initial attempts at this suggest it will be very effective - I've already 
learned of a number of error conditions that it simply would never have 
occurred to me to search for.

The problem is, I don't see how to make this work long-term because it 
requires a rather longer search than Graylog allows for.

Let me elaborate:  I have put together a search that eliminates normal 
behaviour thus:


   1. source:(SERVER NAME)
   2. AND ( NOT ( EventID:[4768 TO 4770] AND _exists_:SourceBlock))  
   3. AND ( NOT ( EventID:4672 AND _exists_:SourceBlock)) 
   4. AND NOT (EventID:4933 AND EventType:AUDIT_SUCCESS)
   5. AND NOT (EventID:14544 AND SMBShare:Shares AND Severity:INFO)
   6. AND NOT (EventID:5312 AND Severity:INFO)
   

This is a highly cut-down version of the search - the real thing is about 
ten times as long!

AFAICT, I can't express syntax like this in a stream or an extractor 
(because neither can work on compound expressions based on two or more 
fields). So I've got to do it as a search.

The problem is, the search is sent as an HTTP GET request, and starts to 
fall over once it exceeds a certain length. I run out of space some time 
before I run out of routine things that I want to exclude from my results!


The most elegant solution I can think of is to write a set of Drools rules 
that encompass all these constraints, then set a field for every message 
that meets this set of rules. This would allow me to make the search as 
simple as "source:(SERVER NAME) AND NOT _exists_:RoutineMessage"


Have I missed something, or is this my best option?



James

-- 
You received this message because you are subscribed to the Google Groups 
"Graylog Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To view this discussion on the web visit 
https://groups.google.com/d/msgid/graylog2/85ae2c58-6163-431f-8b7e-150e9d5c6f28%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Reply via email to