success!

I ran the ossec-remoted in debug mode on the manager and found a
stream of these:

2011/02/26 13:15:48 ossec-remoted(1403): ERROR: Incorrectly formated
message from '192.168.0.1'.
2011/02/26 13:15:53 ossec-remoted(1403): ERROR: Incorrectly formated
message from '192.168.0.1'.
2011/02/26 13:15:59 ossec-remoted(1403): ERROR: Incorrectly formated
message from '192.168.0.1'.
2011/02/26 13:21:43 ossec-remoted(1403): ERROR: Incorrectly formated
message from '192.168.0.1'.
2011/02/26 13:21:49 ossec-remoted(1403): ERROR: Incorrectly formated
message from '192.168.0.1'.

some googling told me that this is sometimes related to the keys for
the agent.  I looked at the client.keys file on both agent and manager
and found that the manager had several entries for the 192.168.0.1 ip
address (while troubleshooting, i removed the agent and re-added it
once or twice).  the deleted entries showed the IP address as
"#######-0.1" or something like that.  I should have kept a copy, but
i just deleted those lines from the client.keys on the manager leaving
only the current key/agent #.  restarted both manager and agent, and
poof, everything started working.

so, lessons:

ossec.conf on agent only needs the <server-ip> section.  everything
else can be done in the agent.conf.
active-responses work when the agent is properly configured and
properly communicating with the manager.
running the manager daemons in debug mode can be very informative.

definetly got to know ossec a bit better these last few days.

cheers,

J

On Fri, Feb 25, 2011 at 9:25 PM, Joel Brooks <[email protected]> wrote:
> i can get the active response to fire by passing "-b 1.2.3.4 -f
> firewall-drop600 -u 000"
>
> firewall-drop600 is in the ar.conf.
>
> I guess i don't (yet) understand what uses ar.conf and what uses ossec.conf.
>
> --------brain dump-------------
> from what i think i understand then,
>
> the ossec.conf on the manager controls that manager config.
>
> the manager decides when an active response should run and simply
> tells an agent to block an ip address by using a script (much like
> calling agent_control -b... by hand).
>
> the ossec.conf on the agent controls the agent config.  It can be as
> simple as the <server-ip> configuration parameter telling the agent to
> get it's config from the manager, or as complex as defining local
> configuration parameters (including active response rules, etc).
>
> the agent.conf on the manager controls the agents' config.  it can
> essentially contain any of the client config parameters.. syscheck,
> rootcheck, etc.  the agent.conf is modified on the manager and
> distributed to the agents on some time based schedule (1-2 hours?)
> unless both manager and agents are restarted.  the agent.conf on an
> agent should be exactly the same as the agent.conf on the manager (and
> can be verified by the md5sum).
>
> -----------------
> I will try in debug mode, and i will make sure i'm firing a rule that
> is level 6 or higher.
>
> thanks for your patience Dan.
>
> J
>
>
>
>
> On Fri, Feb 25, 2011 at 9:02 PM, dan (ddp) <[email protected]> wrote:
>> Hi Joel,
>>
>> On Fri, Feb 25, 2011 at 7:59 PM, Joel Brooks <[email protected]> wrote:
>>> i still haven't got it working.
>>>
>>> I've tried moving the <command> definitions and the <active-response>
>>> sections to the agent.conf, and still no joy.
>>>
>>
>> No joy because the MANAGER doesn't use the agent.conf.
>>
>>> i just can't get active response to work in central management mode.
>>>
>>> I found that executing
>>>
>>> bin/agent_control -b 1.2.3.4 -f firewall-drop -u 000
>>>
>>> from the manager results in the following in the ossec.log on the agent.
>>>
>>> 2011/02/25 19:53:01 ossec-execd: INFO: Active response command not
>>> present: '/var/ossec/active-response/bin/restart-ossec.cmd'. Not using
>>> it on this system.
>>>
>>
>> The restart-ossec.cmd is a Windows AR, this message can be ignored.
>>
>>> 2011/02/25 19:53:01 ossec-execd(1311): ERROR: Invalid command name
>>> 'firewall-drop' provided.
>>>
>>
>> That's strange. I see it in the ossec.conf you sent to the mailing
>> list (or this could be asking for the actual command
>> /var/ossec/active-response/bin/firewall-drop.sh, I can't test to find
>> out at the moment).
>>
>> Can you turn on debugging on the agents? I'm hoping that might help.
>>
>> I don't think the firewall-drop command would be the one to fire,
>> since the host-deny command is used in the first active-response block
>> and they use the same parameters.
>>
>>> any further insights / thoughts would be greatly appreciated!
>>>
>>> J
>>>
>>
>> When you try testing this with SSH, which alert is firing? Your AR
>> configuration requires that it be level 6+.
>> It looks like most of the (single) ssh authentication failure alerts
>> are level 5 or lower.
>>
>

Reply via email to