monitoring unmanaged nodes

[admin_2]

I've asked the network engineers on my team to set up basic node monitoring of some external entities (for example the remote end of a T-1 line or a remotely hosted web server). However although we have both Ciscoworks

2000 and HP OpenView in use, apparently monitoring of nodes isn't possible

with either of these unless you have at least SNMP Read access to the node.

Of course in the examples I provided above, we don't have that access.

It is hard to understand why a tool as rich as HPOV wouldn't provide such

a basic capability. Basically all we need (as a first step anyway) is a basic

"ping test" to see if the node responds (and we recognize this is a simplistic

test, but it's better than nothing until we have time to come up with something

better).

What are you all doing to handle this type of task?

Thanks!

-Bill

[rmushtaq]

Every NMS application including HPOV uses SNMP to manage the devices. That's the fundamental logic behind all the NMS apps. And you must have configured the SNMP Read community strings on the device in order to poll/manage that using HPOV other other apps. If you don't have configured SNMP or don't know the read strings, then you can't do any SNMP management at all on the device.

As for your question on the basic capability of doing the "ping test", you can also do this via SNMP using the PING-MIB. Details and the procedure can be seen at:

http://www.cisco.com/warp/public/477/21.html. But it also requires to have the SNMP Reas/Write strings confifured on the device.

[#3]

Thanks for the response. I could swear I did this in the past with

Cabletron Spectrum (without knowing SNMP Read strings). The

idea was to just have the NMS do a basic "ping" to see if the

node is alive ... even if it's not really "managed" per se.

So what are you all doing in cases where you do need to know the

status of remote nodes (for example the remote end of an Internet T-1)

where you of course don't have SNMP read access to the remote router?

I've seen other packages (e.g. "WhatsUp" comes to mind from years ago)

which allow this basic functionality in addition to the more robust SNMP

management capabilities. It's hard to believe it's not possible with an

advanced package like HPOV :)

There really are cases where you need (or want) to monitor something

of interest (which you have a right to monitor in the "are you up" sense)

but where you don't have SNMP Read capabilities ...

[rmushtaq]

I'm not aware of how Cabletron devices behave with SNMP, but it's a standard procedure that if you want to manage/poll your devices via SNMP, you need the read-only strings at the minimum for polling. If you can poll a Cabletron device without the SNMP strings that I'd say it's a security problem. You don't want any non-trusted user to poll your devices and read the information :-)

[andrea.meconi]

Hello,

I remeber that with IPM you can generate a synthetic traffic to a web server. So you can test DNS lookup, TCP connect and HTTP transaction time.

Look at http://www.cisco.com/univercd/cc/td/doc/product/rtrmgmt/ipmcw2k/cipm23/cipm23ug/measure.htm')">http://www.cisco.com/univercd/cc/td/doc/product/rtrmgmt/ipmcw2k/cipm23/cipm23ug/measure.htm.

However I think OV can only ping a device and present a colored result of the test on map.

Andrea.

[minielsen]

There is no problem for NNM to monitor the status of a node using ICMP. It's actually what it does by default. No need for SNMP if you just want up/down status of the interface.

You need two things: the node must be discovered (i.e. visibel in the map) and managed (which doesn't require SNMP). Then NNM will use ICMP to poll your node with the interval you specified (default 5 or 15 min. dependant of the version of NNM).

Let me know if you still have problems and I'll help you offline.

BR. Michael.

[iswift]

I agree entirely with Michael.

You can ping ie. 'ICMP poll' any device without knowing it's snmp comm string.

Where you stop at, is being able to find anything about the status of the device other than whether the one interface address you specified is responding.

As other respondents have said, HPOV will add a node to the map if it sees the address in an ARP table of a router that it is able to query.

However, as this isn't an OV forum, I will stick to CW2K.

You could add a device manually to the RME inventory using just it's IP address, and then add that device to RME's Availability Monitor.

You will of course get lots of discovery errors and not-respondings for protocols other than ICMP but if all you want to do is ping, then fine.

LMS or RWAN would seem to be an expensive way of pinging - perhaps the Cisco Works for Small Networks may help you but I don't know any details of that product.