4006 dhcp issue, phantom config entries

[ciscokidjt]

I have two Catalyst 4006's in production, and no other switches (Cisco and non-Cisco) are doing this, but users in one area are having to renew their IP's manually (and we do have portfast turned on). Also, I noticed several entries in the config like 'set port channel 3/1-4 298', and several entries like 'set trunk 4/x off dot1q 1-1005', which were not present in the configs originally. Any clues?

 

[mavos]

We are seeing I think something simular,people booting up and not resolving names until we do an IPCONFIG release then a renew. Looking at it this morning, we believe it is on Windows 2000 clients only, and are testing a couple things especially with the DNS client that you can run on a 2000 client. They are getting an IP address, the last one they had from the DHCP server but no name resolution, we believe the card isn't coming, because of some suspicious events in event log are saying this, so it try's to use it's client DNS, the card comes up uses it's last good address, but doesn't try using the DNS server because it thinks it will use the client DNS. Once we release/renew it sees the DNS server and works fine.

Sounds a little related to yours, I wanted to throw this out to see if you are finding the same thing, and to make sure you haven't found anything with the 4000's which we also over Labor Day installed.

I will look for any new developments from you.

 

[Guest_imported]

I am also having DHCP problems with my new 4006. My users are getting the "No domain server available message" the first time they try to logon. If they hit "cancel" and try again, it works. The problem is only associated with Windows 98 machines. Our Windows 2000 machines aren't experiencing the problem. I have yet to try the portfast solution. Let me know if you come across anything. Thanks.

 

[mavos]

I should clarify so that we are all comparing apples to apples or at least no what relates to us. Our 4006 are at the distribution line of the network, we went with a Cisco preferred design with a Core, Distribution and Access layer approach. At the access layer we use 2924's, there are no users directly connected in the 4006's. But it is after we put the 4006's in that we noticed this issue, could be just coincidence but deffinately I don't want to leave no stone unturned. Right now we are leaning toward the client at the problem, we are NT 4 and migrating to 2000.

 

[ciscokidjt]

Our 4006's are at the edge closets. I did find out that our senior engineer was the one who went in and turned off dot1q trunking. This seemse to have cleared up the problem for us.