[c-nsp] ME3600x SNMP items for power supplies disappear when PSU is pulled

[Phil Mayers]

On 13/09/13 16:21, Richard Hartmann wrote:
> Dear all,
>
>
> we are seeing a rather obnoxious issue on ME3600x with IOS 15.3(1)S.
>
> If a power supply breaks down, we can read out that fact via SNMP.
>
> But if we _pull_ the power supply, the SNMP items simply disappear.
> Which, in turn, leads to our monitoring solution dutifully changing
> the item state to "not supported" and ignoring it.

TBH I would expect it to if there's only one PSU inserted. However, I 
see that other platforms e.g. 6500 under IOS 15 leave rows for the empty 
PSU slots:

CISCO-ENVMON-MIB::ciscoEnvMonSupplyStatusDescr.1 = STRING:  Power Supply 
1, WS-CAC-6000W
CISCO-ENVMON-MIB::ciscoEnvMonSupplyStatusDescr.2 = STRING:  Power Supply 
2, empty
CISCO-ENVMON-MIB::ciscoEnvMonSupplyState.1 = INTEGER: normal(1)
CISCO-ENVMON-MIB::ciscoEnvMonSupplyState.2 = INTEGER: notPresent(5)

...so I guess you want that behaviour?

I think it's arguable which is the "right" approach here - I can see 
legit reasons to do both, and I guess it's got to do with BU logic or 
maybe FRU probing capabilities on the ME3600.

> Alternatively, if there's an item for total number of PSUs, total
> power available, redundant power mode or other ways to determine the
> correct amount of PSUs, that would be appreciated, as well.

Maybe see if there's useful stuff under:

CISCO-ENTITY-FRU-CONTROL-MIB::cefcPowerRedundancyOperMode.18 = INTEGER: 
nonRedundant(4)
CISCO-ENTITY-FRU-CONTROL-MIB::cefcPowerNonRedundantReason.18 = INTEGER: 
singleSupply(3)

...as for "number of supplies" - count all rows in the SupplyState table 
with state==normal?