On Fri, Oct 19, 2012 at 8:59 AM, Christopher Decker
<[email protected]> wrote:
> Dan,
>
> I'm off today, so I won't be able to provide the logs until tomorrow.  They 
> simply say that an opened (syscall=2) failed on /logs/ossec.log.
>

I still want to see them. It'll help me troubleshoot your problem.

> Originally I was NFS (v4) sharing that partition to another box that ran log 
> stash.  I have since removed that share but continue to see the same 
> issues--I suspected that to be the root of the issue, but I have since 
> reverted all of those changes and the problem persists.
>

Did the problem exist before this?

> I performed my OSSEC install by running the install.sh script (and had 
> pre-compiled OSSEC).
>

Maybe try a full installation.

Does ossec-monitord continue to run? Have you tried running it in
debug/foreground mode to see what happens?

>
>
>
> On Oct 19, 2012, at 8:39 AM, "dan (ddp)" <[email protected]> wrote:
>
>> On Thu, Oct 18, 2012 at 5:07 PM, Christopher Decker
>> <[email protected]> wrote:
>>> No.  SELinux is turned off.
>>>
>>> Sent from my iPhone
>>>
>>
>> Please provide the auditd logs (I should have asked for that in the
>> first email). Are there any special mount options on the partition
>> your OSSEC installation exists on? What kind of install did you do?
>> Did you just use the install.sh script?
>>
>>> On Oct 18, 2012, at 8:50 AM, "dan (ddp)" <[email protected]> wrote:
>>>
>>>> On Wed, Oct 17, 2012 at 8:09 PM, Christopher Decker
>>>> <[email protected]> wrote:
>>>>> All,
>>>>>
>>>>> I've been running OSSEC 2.6 in my production environment for about a year
>>>>> now.  I recently had to re-install my OSSEC Manager (due to a recap of
>>>>> hardware).  Immediately after the installation I noticed that auditd was
>>>>> spewing logs because the monitord process was unsuccessfully attempting to
>>>>> open /logs/ossec.log.
>>>>>
>>>>> Eventually auditd begins consuming large amounts of CPU time, and then
>>>>> monitord spins up thread after thread to try and keep up; eventually
>>>>> monitord creates 1000+ threads and the server freaks out.
>>>>>
>>>>> Does anyone have any ideas on what the cause could be?  I've never
>>>>> experienced this with OSSEC and I use it heavily.  At this point I'm
>>>>> assuming it is something simple, as I've spent a considerable amount of 
>>>>> time
>>>>> troubleshooting  :)
>>>>>
>>>>> Notes:
>>>>>
>>>>> My ossec.conf is heavily customized but obviously I do not have a
>>>>> <localfile> entry for /var/ossec/logs/ossec.log.
>>>>> I confirmed the permissions on /var/ossec/logs match those of an OSSEC
>>>>> installation not exhibiting this behavior.  The same is true for the
>>>>> ossec.log file itself.
>>>>> OSSEC is able to write to /var/ossec/logs/ossec.log upon start-up, though
>>>>> I'm guessing the process there is not monitord
>>>>> I know OSSEC is chrooted, but for the hell of it I created a symlink from
>>>>> /logs/ossec.log-->/var/ossec/logs/ossec.log.  This was unhelpful.
>>>>> The server acting as the OSSEC Manager is beefy.
>>>>>
>>>>>
>>>>>
>>>>>
>>>>> Thanks,
>>>>> Chris
>>>>
>>>> Is SELinux blocking it?
>

Reply via email to