[Fusioninventory-user] Questions about netinventory & SNMP models

[Guillaume Rousse]

Le 23/05/2012 19:28, David Durieux a écrit :
>
>
>
> Mario Jorge de Sousa Freire<mario.freire at camara.gov.br>  a écrit :
>
>> Hello!
>>
>> 1) I've heard that in the future users will be able to create their own
>> SNMP models. Do you think this feature will be available on next main
>> version (0.83+1.0)?
>
> Oh no, it's a big work to code this, so not available in 0.83+1.0. I haven't date for this, except if somebody want to pay for this :)
A specification would be enough for most people able to read a technical 
document.

[..]
>> 4) Why net inventory is not implemented stepwise: first get standard
>> MIB (by default) and then try to get additional information using SNMP
>> models?
>
> Because oids names not same between 2 mibs, mibs not always ok with firmware... so not good solution
I seriously doubt any firmware difference for any network device would 
change system location variable, or any other similar standard 
property... Rather than current 'identified and certified  hardware and 
firmware combination, you get all available information, otherwise you 
get nothing', they are a whole lot of "best effort" alternative 
approaches to test, such as:
- if there 'cisco' string in system description is likely to be a 
network device
- if this is a cisco device, all other cisco models can be tried until 
something reasonable is found
- etc...

As Mario, I'd prefer to cover 80% of the existing world hardware with 
80% precision, rather than 20% with 100% precision. Especially without 
an authoritative list of those 20%...
-- 
Sex is like snow; you never know how many inches you are going to get or 
how long it is going to last
		-- Murphy's Laws on Sex n°9