Risto,

   Aha, thank you so much, I forgot this. I read all the docs, official 
and not before starting with Sec but forgot about this bit, perhaps I 
read too much documentation...

Thanks again, I love Sec by the way...

-h

Risto Vaarandi wrote:
> Hari,
>
> the problem you have can be tackled in a simple way -- an appropriate value 
> must be chosen for the 'desc' parameter of the SingleWithSuppress rule. Since 
> you mentioned the problem of changing timestamps, I suppose that right now 
> you probably have 'desc=$0' in your rule definition. 
>
> Since the 'desc' parameter is evaluated for each matching event and forms the 
> event correlation key, *and* $0 equals to the entire matching line (including 
> the timestamp!), the event correlation key will be different for each 
> matching event. As a consequence, a new event correlation operation is 
> started for each matching event, and thus no event suppression is done. 
>
> If you would like to suppress repeated invalid user events for the same user 
> name, use the following 'pattern' and 'desc':
>
> pattern=sshd\[\d+\]: Invalid user (\S+) from \S+
> desc=Failed login for user $1
>
> If you would like to suppress such events for all user names for a certain 
> time frame, you could rewrite 'desc' as:
>
> desc=Failed logins for all users
>
> In that way, the event correlation key is constant for all "Invalid user" 
> events, and all such events will be correlated by one operation only.
>
> Finally, the following section of the SEC man page describes the role of the 
> 'desc' field in the event correlation process: 
> http://kodu.neti.ee/~risto/sec/sec.pl.html#lbAT. Have a look at this, because 
> it explains the relationship between rules and event correlation operations.
>
> hth,
> risto
>
>
>   
>> From: Hari Sekhon <[EMAIL PROTECTED]>
>> Subject: [Simple-evcorr-users] Suppression doesn't work if event is slightly 
>> differently timestamped.
>> To: [email protected]
>> Date: Monday, July 7, 2008, 8:14 AM
>> Hi,
>>
>>    I'm using Sec for event correlation and alerting and
>> am finding 
>> myself trying to figure out how to suppress similar but not
>> identical 
>> events.
>>
>> I have used SingleWithSuppress to suppress an event but if
>> the event 
>> text is slightly different I find that it does not really
>> suppress it.
>>
>> eg.
>>
>> Jul  3 12:33:49 hostname sshd[4801]: Invalid user UserXYZ
>> from x.x.x.x
>>
>> will be suppressed but then
>>
>> Jul  3 12:33:56 hostname sshd[4801]: Invalid user UserXYZ
>> from x.x.x.x
>>
>> will not be because the prefix timestamp has changed and
>> made the log 
>> event look different, so the action will apply to this,
>> even if it is 
>> within the suppression period of the first event of the
>> same type.
>>
>> I figure I can add them to a context to store them like
>> that, which I've 
>> done, but I'd also like the ability to threshold or
>> suppress for those 
>> contexts as well as I don't want that context to have
>> too many entries 
>> of the same event.
>>
>> Anybody got any ideas on this?
>>
>> -h
>>
>> -- 
>> Hari Sekhon
>>
>>
>> -------------------------------------------------------------------------
>> Sponsored by: SourceForge.net Community Choice Awards: VOTE
>> NOW!
>> Studies have shown that voting for your favorite open
>> source project,
>> along with a healthy diet, reduces your potential for
>> chronic lameness
>> and boredom. Vote Now at
>> http://www.sourceforge.net/community/cca08
>> _______________________________________________
>> Simple-evcorr-users mailing list
>> [email protected]
>> https://lists.sourceforge.net/lists/listinfo/simple-evcorr-users
>>     
>
>
>       
>
>   


-- 
Hari Sekhon


-------------------------------------------------------------------------
Sponsored by: SourceForge.net Community Choice Awards: VOTE NOW!
Studies have shown that voting for your favorite open source project,
along with a healthy diet, reduces your potential for chronic lameness
and boredom. Vote Now at http://www.sourceforge.net/community/cca08
_______________________________________________
Simple-evcorr-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/simple-evcorr-users

Reply via email to