cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
965
Views
20
Helpful
7
Replies

CS 3.1.1 Device Discovery

s.leyer
Level 1
Level 1

Hello,

We have problems with Catalysts which have several IPs and domain names.

I have 2 questions:

1) During Device Discovery CS doesn't detect that 2 names or 2 IPs belong to the same device. How can I avoid to get the same device twice in the database ?

2) Sometimes, after the discovery process, CS changes the IP-entry of an existing device. This may cause trouble.

Is it possible to fix the IP ?

Regards

Silvia

1 Accepted Solution

Accepted Solutions

Joe Clarke
Cisco Employee
Cisco Employee

This is a known bug, CSCsm66330. There is an experimental patch available by contacting the TAC. The suggested workaround to use cdp source-interface is a good one.

Assuming you choose a name resolution scheme like Use Loopback, you shouldn't have a problem with device names changing. If you use Resolve By Name, then the device name may change depending on the interface first discovered. The cdp source-interface workaround may help this as well.

View solution in original post

7 Replies 7

derek.sizeland
Level 1
Level 1

The best solution is to use the following command:

cdp source-interface 'loopback_ip'

You will need to check that the IOS version on your switches is late enough to support this command.

I have the same problem with some CAT6509 switches running 12.2(33)SXH which doesn't support this command.

Joe Clarke
Cisco Employee
Cisco Employee

This is a known bug, CSCsm66330. There is an experimental patch available by contacting the TAC. The suggested workaround to use cdp source-interface is a good one.

Assuming you choose a name resolution scheme like Use Loopback, you shouldn't have a problem with device names changing. If you use Resolve By Name, then the device name may change depending on the interface first discovered. The cdp source-interface workaround may help this as well.

if 'Use Loopback' is choosen as a name resolution scheme, what happens to devices that do not have a loopback interface configured or even cannot have a loopback interface (like PIX or ASA devices) but have (some) interfaces configured in DNS?

Which interface will be the management interface?

Is the selection criteria for the management interface bound to the ifType (or to what else) ?

Will name resolution mechanism take place even it is not a loopback interface (ifType = 24) or will it be skipped?

The interface identification is tied to ifType 24. However, if no loopback is present, the algorithm falls back to highest IP address of an ethernet interface (ifTypes 6 and 26), then highest IP address of a tokenring interface (ifTypes 8 and 9), then highest IP of a serial interface (ifTypes 22 and 53).

I have a similar Problem with a PIX, where the

outside IP address always gets chosen (it is on a ethernetCsmacd(6)), while the management interface (iftype l2vlan(135)) is ignored.

Why do they give ifType 135 interfaces such a low priority?

I think it should be the same as ifTypes

6 ans 26, or am I missing something here?

Thanks for your replies.

"cdp source-interface" seems to be a good idea. But we mostly have in use Catalysts 6500, 4500 and 3750 and the command doesn't seem to be supported. "show interface loopback" does not work. I've never changed the configuration of a loopback interface and I couldn't find an appropriated document yet. I'll ask for the patch.

The patch solved the problem. I'm glad about it.