Hi Robert,

How many events per second is the manager processing right now?

Such high CPU utilization is not normal.. You can see these values at
/var/ossec/stats . Also, anything special in the manager log when the
disconnect happen?

Thanks,

--
Daniel B. Cid
dcid ( at ) ossec.net

On Thu, Aug 5, 2010 at 7:49 PM, Griffith, Robert
<[email protected]> wrote:
>
>  Attached is a portion of the Client Log that shows the exact time when an 
> OSSEC client disconnects & fails to reconnect.  Also is snswmsta1.pcap which 
> contains the network trace of that particular instance and the communication 
> between both servers.  It shows the  conversation between the client (Source) 
> and the OSSEC server (Destination).
>
> Thank you,
> Robert
>
>
> -----Original Message-----
> From: [email protected] [mailto:[email protected]] On 
> Behalf Of Griffith, Robert
> Sent: Wednesday, July 28, 2010 3:36 PM
> To: '[email protected]'
> Subject: RE: [ossec-list] All UNIX/LINUX agents disconnecting and failing to 
> reconnect
>
>
> Also, on our Ossec Master Server, we are observing that the "ossec-analysisd" 
>  uses ~100% of a single CPU (4 CPU's are available).  Could this be causing 
> any issues?
>
>  PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND
>  2779 ossec     25   0  8236 2932  704 R 93.2  0.1 119:14.70 ossec-analysisd
>
> Thank you,
> Robert
>
>
> -----Original Message-----
> From: [email protected] [mailto:[email protected]] On 
> Behalf Of Griffith, Robert
> Sent: Wednesday, July 28, 2010 5:14 AM
> To: '[email protected]'
> Subject: [ossec-list] All UNIX/LINUX agents disconnecting and failing to 
> reconnect
> Importance: High
>
>  We continue to observe the Ossec Client disconnects to our Ossec Master 
> Server over the network.  We started receiving these disconnects again on 
> Wednesday July 7th 2010.  The clients disconnected daily and failed to 
> reconnect for hours (some clients took days or never reconnected again).  
> This  issue was also observed in May on 5/10/2010.  That issue lasted for two 
> weeks and then suddenly stopped without any Ossec configuration changes.
>
>  We implemented the fix you provided below after we encountered the issue 
> again on Wednesday July 7th 2010.  We cleaned the "rids" directory and 
> disabled the counters on both the Ossec Master Server and all UNIX/Linux 
> Clients (set verify_msg_id to 0 on the internal_options.conf).  The "fix" you 
> provided cleared the issue for 5 days and then the Client disconnect issue 
> re-emerged.  We re-applied the "fix" again without success.
>
>  We are again experiencing disconnects and failed re-connects on all 
> UNIX/LINUX Ossec agents.
>
> FYI: We are using Ossec Version 2.4.  Counters are disabled.
>
> Thank you,
> Robert
>
>
> -----Original Message-----
> From: [email protected] [mailto:[email protected]] On 
> Behalf Of Daniel Cid
> Sent: Friday, May 14, 2010 9:43 AM
> To: [email protected]
> Subject: Re: [ossec-list] RE: All UNIX/LINUX agents disconnecting
>
> Hi Lucio,
>
> There is two issues in this thread. One, the agent disconnects and then 
> reconnects by itself. That's fine and can happen on high load environment or 
> when a message gets dropped.
>
> The second issue that Mike mentioned happens when the counters get out of 
> sync and the agent never reconnects. For this problem, you have to either 
> clean the "rids" directory on the manager or disable the counters. To disable 
> it, set verify_msg_id to 0 on the internal_options.conf file:
>
> # Verify msg id (set to 0 to disable it) remoted.verify_msg_id=0
>
> Thanks,
>
> --
> Daniel B. Cid
> dcid ( at ) ossec.net
>
>
> On Thu, May 13, 2010 at 1:21 PM, Lucio Emanuel Soldo <[email protected]> 
> wrote:
>> Hi Mike, how are you? Could you post the final solution your team has
>> produced in order to fix its problem?
>>
>> Thanx alot!
>>
>> On Tue, May 11, 2010 at 6:56 PM, Pendergrast, Michael L
>> <[email protected]> wrote:
>>>
>>> Yes we have
>>>
>>> although we have v1.6
>>>
>>> I don't have the details as my team has worked the problem and is
>>> currently deployed.
>>>
>>> What we did find is that there is a counter in the agent and in the
>>> manager and if they get out of sequence the agent will stop
>>> (basicaqlly they get out of sequence).  We also found that at startup
>>> of the UNIX agents that if multiple agents all start at the same
>>> time, the agents will stop.  In this case, for initial startup we had
>>> to sequence the startup in about 10 min increments.
>>>
>>> Mike
>>> ________________________________
>>> From: [email protected]
>>> [mailto:[email protected]] On Behalf Of Griffith, Robert
>>> Sent: Tuesday, May 11, 2010 12:26 PM
>>> To: '[email protected]'
>>> Subject: [ossec-list] All UNIX/LINUX agents disconnecting
>>> Importance: High
>>>
>>>   We have been running the new version of Ossec 2.4 in our
>>> environment for
>>> 3 weeks.  Yesterday all of our UNIX/LINUX client agents started
>>> disconnecting.  None of our Windows Server client agents have disconnected.
>>> Has anyone experienced this and/or found a resolution for this issue.
>>>
>>> Thank you,
>>> Robert
>>>
>>
>

Reply via email to