On Thursday 06 April 2006 10:47, Steve Brueckner wrote:
What might cause this?
The event ID can be recycled. Its the combination of time stamp and serial
number that creates uniqueness.
At some point my event IDs got reset (they didn't cycle that
fast!). I've
been playing quite a bit with the audit system so I'm not sure what caused
it. Possibilities include:
- Restarting the auditd service
Nope
- Rebooting the machine
Yep
- Deleting the /var/log/audit/audit.log file
Nope
There can also be wrapping.
Or should this just plain not happen?
It can happen.
I'm on FC4 using kernel 2.6.12-1.1447_FC4xen0. I'm afraid I
can't easily
upgrade at the moment because I've build an entire system predicated on old
versions of SELinux and Xen.
You will likely have other problems on a kernel that old. I think 2.6.14 was
when we really had most features in place and stable.
I also have a couple of other questions:
- How large to audit event numbers get before they cycle back to zero?
I think its a u32 number.
- Is there any way to have ausearch only the most recent audit log
instead
of all logs?
Sure, use the "-if" option and give it the full path to the file.
ausearch -if /var/log/audit/audit.log
-Steve