Nagios Agent mode

[Thomas Guyot-Sionnest]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 04/04/08 03:56 PM, Steve Pribyl wrote:
> Good Afternoon,

Hi Steve,

You contacted me in the past about OCP_Daemon... Is there anything in it
that didn't fit your needs? I'd be glad to hear you opinion...

> Has there been any interest in building a agent mode for nagios.  Seems
> like most of the pieces are already there they just need to be put
> together.
> 
> The nagios server would be configured to monitor the host and services
> passively and nsca would be installed an running.
> 
> The nagios agent would monitor the services local and "ping" the server
> for host alive.  An important difference would be only changes in state
> would be sent to the server using code from send_nsca.

I'm not sure if I fully understand what you'd like but this reminds me
how MOM operates (M$ Operation Manager), which is, from my experience, a
piece of sh*t product... The main flaw can be resumed on one sarcastic
phrase: It waits for the server to notify MOM it's dead*.

> I am aware of nrpe but it does not scale well in that it generates lost of
> network traffic and server load as the network grows.
> I am also aware of the OCP_Deamon trick, however it sends everything and
> interferes with perfdata.
> I have tried a using notifications with send_nsca instead of mail with
> little success.
> I event developed a POC check_nsca plugin, that wraps other plugins and
> forwards them to the server, similar to check_nrpe.
> 
> I am very interested in taking this up if there is interest.  I have
> experience in C,C++ and writing systems monitoring software on oodles of
> platforms(I can elaborate off list for those interested).

Well, I had in mind something a bit more traditional that would likely
help a bit... I will explain a bit more in detail later (stay posted on
nagiosplug-devel/users lists) but roughly I'm thinking we could write an
SNMP subagent that would "compiles in" the Nagios plugins and allow
running them trough the snmp daemon. The snmp protocol is very
lightweight and since the plugin code would live right in the subagent
it tould not even require forking...

This is a bit ambitious (apart from requiring strict memory handling
from the plugins, it would likely need caching and queuing) so
additional resources would likely be required. OTOH I have enough SNMP
knowledge to write the MIBs and work on the agent code.


Thomas
- --
* Not actually true as there's heartbeats, but in countless cases it
never detected the source of the problem despite the fact that it was
"monitored".
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFH+FR56dZ+Kt5BchYRAhWRAKCcCBcmUxdTv1jkBGpNqeHW88sAFACgve6Q
ihnOyEkaYhC9x6tJGtl9BQw=
=Fvbt
-----END PGP SIGNATURE-----

-------------------------------------------------------------------------
This SF.net email is sponsored by the 2008 JavaOne(SM) Conference 
Register now and save $200. Hurry, offer ends at 11:59 p.m., 
Monday, April 7! Use priority code J8TLD2. 
http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone