How to filter SYSLOG data to collect meaningful information only?

Dear users,

SUSE 10 sp3 and SUSE 11.

I made configuration changes in '/etc/syslog-ng/syslog-ng.conf'
to move SYSLOG content to LogRhythm.

This is what I changed in the file.
Un-comment out the following lines
#Enable this and admop IP to send log messages to a log server
Destination logserver - Enter in syslog server ip address
log {source(src)
destination allmessages

Now, logs are being generated and transferred to LogRhythm. But the problem is too much log information is being generated and it's filling up our Logrhythm quickly, 30% in a week. We'd like to reduce log generated in SYSLOG by collecting only useful information which describe below.

The log information we'd like to capture is below.

what file should I edit and what changes should I make in that file? Is there any document or procedure posted on the web?

Always appreciate your support.

  • Successful login and logoff attempts

  • Unsuccessful login and authorization attempts

  • All identification and authentication attempts

  • All actions, connections and requests performed by privileged users

  • All changes to logical access control authorities (e.g., rights, permissions

  • System changes with the potential to compromise the integrity of audit -policy configurations, security policy configurations and audit record generation services.

  • Creation, modification and deletion of objects including files, directories and user accounts

  • Creation, modification and deletion of user accounts and group accounts

  • Creation, modification and deletion of user account and group account privileges

  • The date of the system event; ii)the time of the system event; iii) the type of system event initiated; and iv) the user account, system account, service or process responsible for initiating the system event.

  • System start-up and shutdown functions.

  • Modifications to administrator account(s) and administrator group
    account(s)including: i) escalation of user account privileges
    commensurate with administrator-equivalent account(s); and ii) adding or
    deleting users from the administrator group account(s).

enabling or disabling of audit report generation services

-command line changes, batch file changes and queries made to the system (e.g., operating system, application, and database).

I would continue to send all logging data to your logging server and use a tool like Solarwinds LEM (Log and Event Manager) to produce the reports and eventing output you looking for.

is there a way to control producing or generating SYSLOG data?

I'd like SYSLOG to generate only certain data such as update, delete, logon etc.

We have LogRhythm and it doesn't want to store all SYSLOG data generated.

Thanks,

The SYSLOG file '/var/log/messages/' generates so many lines of Mcafee scan results.

I'd like to get rid of these lines.

The log lines start with 'nails-ddds' and end with 'cmd=update'.

What command do I add in '/etc/syslog-ng/syslog-ng.conf' file to get rid of these log information?

Thank you,

You need to consult your McAfee manual and see if the "facility" is configurable. Usually, you'd choose one of the "user" facility types (local0 - local7) and configure it to use that facility... then you can adjust your syslog conf to take messages for that facility and output to a separate log area.

Alternatively, McAfee may have support for outputting to a local log file outside of using syslog messages, in which case you can configure that.

Syslog really doesn't support the idea of fine grained log selection (which is why SolarWinds and Splunk and such exist).

With that said, as systemd (the borg) takes over everything, logging is going to change in some pretty radical ways... so whatever you do with syslog today, don't get used to it. I promise you it will change if your distribution switches to systemd.

You never know what you'll want from your logfiles until you do, so I wouldn't try being too incredibly specific except for separating mcaffee results from everything else interesting.

I would go for collecting everything and then extracting what you actually need when you need it. It leaves masses of records never read, but there is no way to say to the server, "Hey, just go back in time, replay a process that I don't know quite when it happened and show me the logging now."

We have load and load & LOADS of output that is never looked at unless there is a problem and we rotate the logs out and away at sensible intervals to keep on top of it. if we don't need most of it in a week (production batch messages) then they are sent to tape and never seen again.

Robin

You said, "You never know what you'll want from your logfiles until you do, so I wouldn't try being too incredibly specific except for separating mcaffee results from everything else interesting.".

How do you separate McAfee results from everything else?

Thanks,

First I would check the suggestion cjcox made last week:

In Suse Linux servers, there is a syslog file /var/log/messages The file size is 5G and it's getting filled up with logs.

What happens when it reaches to full? Is the system going to stop? Is there a mechanism to clean up or rotate automatically so that the file never reaches full?

Thanks,

Post this as a new entry please.