Re: [Devmon] different templates for hardware reporting identical sysdesc

[Buchan M. <bg...@st...>]

On Tuesday, 27 September 2011 05:28:37 Jeremy Laidman wrote:
 
> What would be really useful is to be able to specify the device matching in
> the spec file more precisely.

Well, sysDescr use isn't a great idea, and I was considering using 
sysObjectId, as that will (at least for devices that support it) be the 
easiest to use to identify the device without requiring access to all the 
devices for which templates exist.

However, this is only sufficient to identify the base device/appliance, not 
any optional features it may support (e.g. feature blades on modular chassis 
such as a load balancer in a Cisco 6500/7600 or more modern replacements).

sysORID may help somewhat here to identify standards-based MIBs the device 
supports, but not any vendor-specific MIBs.

Some of this work may be done more efficiently with better MIB support, and 
requires changes to how templates are handled, so I think some more 
experimentation needs to be done here, and most likely in a new branch or 
release series.

So, I would really like to get the 0.3.1 release out of the way before 
starting work on any of these issues.

> For example, it would be neat to be able to
> add one of these into the spec file:
> 
> oid-exists : 1.3.6.1.4.1.232.3.2.4.1.1.1
> oid-exists : tcpConnLocalPort.127.0.0.1.25.0.0.0.0.0

To identify a server with a running MTA? What would the value here be?

> oid-value : 1.3.6.1.4.1.232.3.2.4.1.1.1 > 0
> oid-value : sysLocation =~ /Site2/
> 
> (The OID used here refers to the HP/Compaq drive hot-spare count.)

Yes, I guess that could be useful to distinguish the hot-spare vs no-hot-spare 
case. But, I don't know if separate templates is the best fix for this issue, 
or if instead empty tables should be ignored totally.

I will try and look at solutions for this problem in October or November.

Regards,
Buchan