Philippe, by "managing and using" the audit logs after centralizing, what do you mean?

You can centralize logfiles using rsyslog to a "central" (r)syslog server, then you can find the logs in whatever path you decided to store them and use the ausearch command and seek out whatever records you want based on audit record details you can query.

It is important though to fully and discretely articulate what you mean by "managing and using" audit logs after centralizing.
Perhaps what you mean by managing is:  logrotation?
Perhaps what you mean by using is: query audit-record data based on specific filter-capable details?


--------------------------
Warron French



On Fri, Feb 7, 2020 at 2:15 PM Steve Grubb <sgrubb@redhat.com> wrote:
On Friday, February 7, 2020 5:02:01 AM EST MAUPERTUIS, PHILIPPE wrote:
> Apart the man pages, I didn’t find anything useful relating to
> audisp-remote. I am searching information on how it scales ? Is there any
> performance issue ? How to use it in a large environment ?

It is really designed for smaller deployments. If you have 10 or so systems,
it should do OK. I have not tested load handling of the daemon via network
sources. But have tested the ability to write logs and its very fast. Fast
enough to fill your hard drive in a minute or so.

> Most of what I found dated a long time ago and mainly said use rsyslog
> instead. It seems that centralizing the messages through rsyslog is far
> more popular. Is audisp-remote really used ?

For small deployments sure. If you really have a lot, then you probably
should use some kind of subsystem designed to handle large amounts of data.
ELK, graylog, splunk are all a couple that come to mind. I also suspect you
want audit data correlated with other application information.

The main issues at scale are log management, searching, and alerting. These
are all problems that one person hacking on spare time can't really achieve
well. If we had a stronger community with more participants, we probably
would have better and nicer tools.

> The man page read :
>        tcp_max_per_addr
>               This  is  a  numeric  value  which  indicates  how many
> concurrent connections from one IP address is allowed.  The default is 1
> and the maximum is 1024. Setting this too large may allow for a Denial of
> Service attack on  the  log‐ ging  server.  Also  note  that the kernel
> has an internal maximum that will eventually prevent this even if auditd
> allows it by config. The default should be adequate in most cases unless a
> custom written recovery script  runs  to forward unsent events. In this
> case you would increase the number only large enough to let it in too.
> Where could I find an example of recovery script ?
> Could it be a way to inject the audit message in auditd after having
> receiving them via rsyslog ? This might be useful just because, by default
> ausearch in all available logs and the -if parameter accepts only one
> file.

I think you can inject logs by

ausearch --start XXX --raw | audisp-remote


> Maybe my lack of knowledge about auditd leads me to write rubbish.
> If so, please direct me to where I can find how to manage and use audit
> logs after centralizing them. Not only keeping them but  acutually using
> them.

There may be others in the community that can offer some insight here.

-Steve



--
Linux-audit mailing list
Linux-audit@redhat.com
https://www.redhat.com/mailman/listinfo/linux-audit