Yes, you're correct, responding correctly to stimulus is all that is needed, there's nothing that mandates python.
Beyond that, I'm no help, I failed to figure out how to successfully model devices via ssh when I originally setup my zenoss monitoring box long ago. I know it can be done, but I never figured it out... -Scott On Mon, Dec 1, 2008 at 9:23 PM, mdepriest <[EMAIL PROTECTED]> wrote: > (Sorry for the cross post but it looks like there's a 10:1 traffic > difference between the zenoss-users and zenoss-dev forums.) > > I have searched far and wide, read the Developer's Guide and the Admin > Guide, and walked the wiki. There's a lot of great info in those places but > I'm still struggling with what is probably a basic concept regarding > modeling without SNMP. I have tried the various instructions regarding > setting up SSH Command based modeling and I still come up short in > understanding what I need to know to solve my problem. > > I need to be able to model and monitor devices which do not support SNMP. > The monitoring part I have come to grips with, thanks to the forum and the > suggestion to use Nagios plugins to collect data. What I don't know how to > do is model the devices in the first place. It's all great to be able to > report on file system usage, but unless you can create a list of filesystems > to monitor, it is of moderate utility. > > I am able to configure a device to use SSH Command modeling as per the > Admin Guide. However, running the modeler gets me nothing for that device. > > Unless I totally misunderstand the process, SSH Command style modeling > depends upon an "agent" (the Zenoss Plugins) residing on the system to be > modeled. The Zenoss Plugins do a wonderful job, but only if you satisfy two > basic constraints: a) your device is on a supported platform like Linux, and > b) your device can run Python scripts. > > My situation is that the devices I want to model are neither on the > supported list or capable of installing Python. They can, however, be > connected to via SSH and run scripts that can take arguments and return > output on stdout. It would seem to me that in the end that is all that is > required - that the remote plugin doesn't actually have to be in Python, > just respond to the correct stimulus in the correct way. > > So, I reason that it should be possible for me to script whatever I want. > However, I can't find anywhere to confirm that reasoning, nor can I find any > documentation on what the acceptable format for the response should be. > (That doesn't mean it's not there, it just means I didn't find it in my > searching.) > > Am I correct? Or is there some voodoo that requires Python on the device to > make the interface work? And where can I learn more about what the content > of the plugin has to be like, and what the handshake between systems is > during modeling? > > Thanks in advance > > --Mike > > > > > -------------------- m2f -------------------- > > Read this topic online here: > http://forums.zenoss.com/viewtopic.php?p=28490#28490 > > -------------------- m2f -------------------- > > > > _______________________________________________ > zenoss-users mailing list > [email protected] > http://lists.zenoss.org/mailman/listinfo/zenoss-users >
_______________________________________________ zenoss-users mailing list [email protected] http://lists.zenoss.org/mailman/listinfo/zenoss-users
