Hi,

I did some testing and it seems that anytime a log file
changes, the analysis engine is triggered. So monitoring log
files seems a rather expensive operation (and very useful of
course). In the scenario of several virtual servers running
on a real host, I would setup an ossec server installation
on the real host. On the virtual servers will be only ossec
agent for for rootkit checking and nothing more. This way, I
hope I don't miss anything that ossec offers while the
performace cost is reasonable. Any comments/suggestions
please?

Thanks,
Thanh


On Sun, Mar 18, 2007 at 10:49:15AM +0100, The Thanh Han wrote:
Hi Daniel,

many thanks for the references. It's clear to me now that the
ossec agent must run inside each virtual server to be able
to detect rootkit.

Do you think it's better to let the real host monitor logs
and file integrity of the virtual servers? Or it's better to
have the agents doing these work and send alert to the
server (the real host)? The first variant seems to me
slightly more efficient, but I am not sure since.
Another advantage of the first variant is that if a virtual
server is compromised, the intruder doesn't know that the
system is being monitored.

Is it safe to commment out rule sets that are not relevant
to a particular system? For example if I am not running
apache on my system, then commenting out apache_rules.xml
is a good (and safe) thing to do?

btw, how often are the log files checked? Everytime when
syslogd is active?

Thanks,
Thanh


On Sun, Mar 18, 2007 at 12:11:52AM -0400, Daniel Cid wrote:
> Hi Thanh,
>
> Your assumtpions are right. You just need to add the files to be monitored
> and the logs to be analyzed. Rootkit detection will only fully work
> for the root server (not the virtual ones), unless you install ossec
> on each virtual server.
>
> The following links can help (regarding rootcheck):
>
> http://www.ossec.net/wiki/index.php/Know_How:Rootkit_Detection
> http://www.ossec.net/dcid/?p=25
>
> Daniel
>
> On 3/17/07, Thanh Han The <[EMAIL PROTECTED]> wrote:
> >
> >Hi Daniel,
> >
> >many thanks for the great reply. I have been playing with
> >ossec and like it very much.
> >
> >Another question is that if I run a virtual server (using
> >vserver or openvz) whose root is ie /var/myserver and would
> >like ossec to protect that virtual server from the real
> >host, then which steps are needed? Probably I have to tell
> >ossec which extra log files to check
> >(/var/myserver/var/log/...) and which files to monitor
> >integrity (like /var/myserver/{/bin,/etc,/sbin,...}), but
> >what about rootkit check? Is there some doc about how
> >rootcheck works? I took a look at the rootkit_files.txt
> >file, but didn't get a clue.
> >
> >Best regards,
> >Thanh
> >
> >
> >
> >On Fri, Mar 16, 2007 at 11:23:45PM -0400, Daniel Cid wrote:
> >> Hi Thanh,
> >>
> >> Currently there is no "official" way to do what you want. You could hack
> >the
> >> ossec2db script (from Meir) for instead of inserting into a db, to
> >generate
> >> the desired e-mail message. In the future, I plan to add support for SMS
> >> specific messages and some additional alerting options, but that will be
> >> in a future version (1.2 and above)..
> >>
> >> *Btw, the current ossec-maild works fine with gmail SMTP (I used it all
> >> the time), since you are not required to use TLS for it.
> >>
> >> Thanks,
> >>
> >> --
> >> Daniel B. Cid
> >> dcid ( at ) ossec.net
> >>
> >> On 3/15/07, Thanh Han The <[EMAIL PROTECTED]> wrote:
> >> >
> >> >Hi list,
> >> >
> >> >is it possible to use some other program to send mail alert,
> >> >instead of ossec-maild? For example, if I want to send
> >> >mail alert to a google account, then SMTP authentication via
> >> >TLS is required and I cannot figure out how to do that. Any
> >> >hint please?
> >> >
> >> >Thanks,
> >> >Thanh
> >> >
> >

Reply via email to