Cisco Support Community
cancel
Showing results for 
Search instead for 
Did you mean: 
Announcements
Webcast-Catalyst9k

Clearing VLAN 1 - effects on CDP, UDLD, PAgP, DTP, VTP

I am confused about what happens when you clear and/or prune VLAN 1 from a trunk. The literature says that you can clear VLAN 1 from a trunk, and that will stop VLAN 1 user traffic and BPDUs from traversing the trunk. However, management protocols such as CDP, PAgP, VTP, and DTP will still pass on VLAN 1, presumably because these protocols are all link-local in their scope.

Now, if these protocols are passing on VLAN 1, what happens if the native VLAN is some other value? Does it mean that those protocols are tagged with VLAN 1? Or are they untagged, and pass effectively on the native VLAN? It is difficult to imagine how DTP could be tagged for example; it's a chicken-and-egg situation. Does it make sense for CDP to be tagged?

Can anyone point me to a protocol specification for DTP? I am interested in frame formats etc. Also for PAgP and UDLD. CDP and VTP seem to be well documented by comparison.

Thanks in advnace.

Kevin Dorrell

Luxembourg

3 REPLIES

Re: Clearing VLAN 1 - effects on CDP, UDLD, PAgP, DTP, VTP

Anyone?

Re: Clearing VLAN 1 - effects on CDP, UDLD, PAgP, DTP, VTP

Kevin

If you have the following config:

interface FastEtherent0/1

switchport access vlan 4000

switchport trunk native vlan 4000

switchport trunk allowed vlan 10,100

switchport mode trunk

CDP, DTP, PaGP will all still be tagged with a VLAN ID of 1. There were some bugs in older 2950/3550 IOS where if the Native VLAN was changed from the default VLAN 1 to anything else then it broke everything. But this was quite a while ago.

It is easy to test by setting up one of your trunk ports as a SPAN source.

Updated. UDLD however is tagged with the Native VLAN ID...... Or at least between my 2950 & 3550 it is.

HTH

Andy

Re: Clearing VLAN 1 - effects on CDP, UDLD, PAgP, DTP, VTP

Hi Andy,

Thanks for the response. I had my doubts about DTP, because it is that which forms the trunk in the first place. So are you saying that while the link is still an access port, it does not tag, but then as soon as it succeeds in negotiating the trunk, it starts to tag the frames?

Also, I was in doubt about whether it was worth trying to SPAN the port to observe the traffic. I know the SPAN can be a bit opaque, for example it does not monitor BDDUs, and I wasn't sure how that applied to the other negotiation protocols.

I suppose that these protocols, being link-local, are not subject to the Spanning-Tree topology, except that maybe VTP?

As for the bug, I came across one recently where clearing the non-1 native VLAN from the trunks caused the VLAN 1 STP to shut down. That one is published for IOS, and has been fixed in recent versions. But it still exists even in the latest version of CatOS on the 4000 series. Melted down my network good and proper!

Any ideas where I can find the protocol specs for all these protocols? The documentation web site somehow does not seem as rich as it was once. And Clarke/Hamilton does not really go into enough detail.

Thanks again.

Kevin Dorrell

Luxembourg

1271
Views
4
Helpful
3
Replies
CreatePlease to create content