NeDi Software Specific > Definition Files

Problem with detecting Duplex mode with some Ciscos

(1/2) > >>


I'm not sure if this is the right place to post this or if Discovery would fit better.

I just had an issue with inconsistent duplex settings between server and switch.
Hence I wonder if I can't use NeDi for detecting more of these beside looking for increasing errors on interfaces.
So I looked around a little bit and fond the following:
The OID used by NeDi for newer Ciscos for determine the duplex settings is:
It returns the setting of the interface (1=HD, 2=FD, 4=Auto or ? in NeDi) but not the actual value.
The OID from the standard dot3 MIB
returns the actual duplex value even if the port is in Auto mode.
But if I put this into the def file, the database row isn't populated.
Maybe it is because of the output of the snmpwalk

--- Code: ---EtherLike-MIB::dot3StatsDuplexStatus.10001 = INTEGER: unknown(1)
EtherLike-MIB::dot3StatsDuplexStatus.10002 = INTEGER: fullDuplex(3)
EtherLike-MIB::dot3StatsDuplexStatus.10003 = INTEGER: fullDuplex(3)
EtherLike-MIB::dot3StatsDuplexStatus.10004 = INTEGER: unknown(1)
EtherLike-MIB::dot3StatsDuplexStatus.10005 = INTEGER: unknown(1)
EtherLike-MIB::dot3StatsDuplexStatus.10006 = INTEGER: fullDuplex(3)
EtherLike-MIB::dot3StatsDuplexStatus.10007 = INTEGER: fullDuplex(3)
EtherLike-MIB::dot3StatsDuplexStatus.10008 = INTEGER: fullDuplex(3)
EtherLike-MIB::dot3StatsDuplexStatus.10009 = INTEGER: unknown(1)

--- End code ---
The value after "INTEGER:" doesn't look like real integer at all  ;)

Any ideas?


The value of "INTEGER" is what snmpwalk translates the integer value to be, based om mib information.
snmpwalk has an option in which the output is not translated, and then integers are shown

On my Ciscos, the Duplex value is displayed correct in Nedi except for ports configured for autosense that are not connected. Those are marked as ? which seems logical to me, they are not full and not half duplex.
I use the DOT3 mib (OID posted above) and it's working correct for the following models: WS-C3560-48PS (and all other 3560 variants we have), Catalyst 6500-IOS (Sup720), WS-C3750


OK, was my own mistake. Changed the OID for IFdupl but not the one for IFduix  :(
It's working now.

Maybe all .def files for 3560 and 2960 should be modified like this in the next release?

Cisco seems to move towards standard mibs, which is a good development :)

The 'ix OIDs should only be populated, if they're not consistent with the ifidx. Otherwise unnecessary walks are performed, which will just slow down the discovery.

If someone can confirm which devs can use standard OIDs (and check, whether 'ix is really needed), I'll gladly update the .defs.


[0] Message Index

[#] Next page

Go to full version