STP with mixed network of Catalyst 3550 and 3500XL

[ghent]

I'm experiencing an issue which seems to be a compatitbility issue with the STP on Catalyst 3550/3750 and 3500XL.

The 3550/3750 switches run (pvst) seperate instances of STP for each VLAN whereas the 3500XL switches run a single STP instance for the entire network.

Therefore these do not operate in the same spanning tree and I got two root bridges, one for each protocol and a loop, it seems.

Any ideas?

Sigurbjartur

[Prashanth Krishnappa]

I am not sure what you mean by "3500XL switches run a single STP instance for the entire network".

2900/3500 XLs do PVST. How many VLANs do you have(Catalyst 3500 XL switches support 64 STP instances and 250 VLANs)? If you are using VTP, do you have more than one VTP domain?

[ghent]

Hi, thanks for your reply.

I'm running less than 64 VLAN in a single VTP domain.

Regarding the 3500XL question. When I do 'sh sp', I get the following response:

"

Spanning tree 1 is executing the IEEE compatible Spanning Tree protocol

Bridge Identifier has priority 32768, address 0004.2775.7b00

Configured hello time 2, max age 20, forward delay 15

Current root has priority 32768, address 0003.6bc1.ab40

Root port is 75, cost of root path is 20

Topology change flag not set, detected flag not set, changes 22

Times: hold 1, topology change 35, notification 2

hello 2, max age 20, forward delay 15

Timers: hello 0, topology change 0, notification 0

"

I do not get a reply per vlan as I do on the 3550.

Could it be because I have an 'allowed vlan' list on the trunk interfaces on the root bridge? I do permit most vlans, but not all.

Sigurbjartur

[Prashanth Krishnappa · Accepted solution]

This means it is PVST. Xls do PVST by default. I would not expect Pruning VLANs to cause the issue you are seeing, however BPDU filtering/Multiple VTP domains can cause it.

If you can, modify your allow list to see if it makes any difference.

[ghent]

That worked, as soon as I stopped pruning and reset the 3550 as root, the switches started converging their links towards it.

When I did some tests it seems that when I prune VLAN 1 on the links of the 3550, which I had done, the neighbor switch stops receiving the BPDUs from it. When I added it to the allowed list, the neighbor started receiving BPDUs again.

This is odd behaviour, considering that I'm not supposed to be able to prune VLAN 1. CDP and VTP worked fine before.

Thank you very much for your assistance.

Regards,

Sigurbjartur

[milan.kulik]

Hi,

just to clarify:

1) There is a difference between VLAN "pruning" and "disabling".

When you enable VLAN pruning on your VTP server, STP BPDUs are still sent in the pruned VLANs, only broadcasts and unknown unicasts are not flooded into pruned VLANs.

When you disable (not allow) VLANx on a trunk port, neither user data nor STP BPDUs are sent to the VLANx on the trunk.

2) VLAN1 is a special case.

When disabled, control protocols (CDP, VTP, PAgP, DTP) should be sent in VLAN1 on a 802.1q trunk.

But not STP BPDUs nor user data.

BUT I remember a 3550 bug CSCdz22629 a year ago which required VLAN1 enabled on a trunk for VTP and CDP to work correctly.

Which IOS version are you running?

Regards,

Milan

[konigl]

What you're seeing when you do "sh sp" is an inconsistency in the switch IOS between the 3550 and the 3500XL. To get a reply per VLAN as you do on the 3550, on the 3500XL enter the command

show spanning-tree brief

The "brief" option does not exist on the 3550.

When you just enter "sh sp" on a 3500XL, it responds as though you entered "show spanning-tree vlan 1".

[ghent]

You seem to have hit the nail on the head and corrected my misunderstanding.

I did not realize that I was only getting VLAN 1 info. by issuing the 'show spanning-tree' command on the XL.

While I was pruning VLAN1 on the links, on my backbone switch, the switches in the VTP domain that were beyond those links, elected a new root bridge for the VLAN among themselves.

Me not realizing that the XLs were running PVST and thought they only ran a single spanning tree instance for the entire network I believed they did not receive BPDUs from the 3550 switch and were sending all their traffic in the wrong direction.

After reading your messages I started checking the other VLANs and noticed that, for those VLANs that are not being pruned in the 3550, STP does indeed have a correct root bridge while a root bridge is being elected for the other VLANs.

Thank you very much for your input.

Regards,

Sigurbjartur

[aashish.c]

Hi,

a simple tip to check which switch is running what mode of STP, you can also cehck it in "sh run".

In the beginning of sh run you will find statement like "spanning-tree mode pvst"

it means that switch is running 1 instance of STP per vlan.

regards

aashish C