Have I got a STP problem?

[Daran Blackwell]

Hello. Any help appreciated with this problem ...

We had a user complaining about a temporary dropped network connection. Looking at the log on the switch the host PC connects to I can see the dropped connection, and it was for a short time (about 20 secs). Now we checked the copper cabling end to end and all was ok, and previously we had a x-over'd uplink flapping on the same switch stack (x-over connections stack, not stackwise). But the logs do not show uplinks to be going down now.

Worryingly I am seeing in the logs that on a daily basis we have the same kind of brief lost connections on 3 or 4 random access ports on every switch in the chain. From the log date / time stamp + the fact the connection drop is so brief it doesn't appear to be PC reboots, I can only think it's a spanning tree issue? The logs show this happening on 1st switch in chain all the way through (this is all happening on a very old legacy leg of Cisco infrastructure btw which is due to be replaced but not sure when).

So hardware setup is:

6500 Chassis

Fiber connection from chassis to port Fa0/26 on 2950 #1 (which has a x-over to another 2950 in same cabinet)

Fibre from port Fa0/25 of that same 2950 #1 to a 2950 in another cabinet (which has another 2 x 2950's off it in a chain via x-over's)

The common log entry I'm seeing daily for random ports on all the switches is this:

Sep 10 08:37:08.203 GDT: %LINEPROTO-5-UPDOWN: Line protocol on Interface FastEthernet0/3, changed state to down

Sep 10 08:37:09.211 GDT: %LINK-3-UPDOWN: Interface FastEthernet0/3, changed state to down

Sep 10 08:37:12.003 GDT: %LINK-3-UPDOWN: Interface FastEthernet0/3, changed state to up

Sep 10 08:37:14.023 GDT: %LINEPROTO-5-UPDOWN: Line protocol on Interface FastEthernet0/3, changed state to up

Sep 10 08:37:25.603 GDT: %LINEPROTO-5-UPDOWN: Line protocol on Interface FastEthernet0/3, changed state to down

Sep 10 08:37:30.223 GDT: %LINEPROTO-5-UPDOWN: Line protocol on Interface FastEthernet0/3, changed state to up

The common STP configuration on all the switches is:

spanning-tree mode pvst

spanning-tree portfast default

no spanning-tree optimize bpdu transmission

spanning-tree extend system-id

And we have 'spanning-tree portfast disable' command on Fa0/1 & 0/2 and on Fa0/25 & 0/26 on all switches.

There is only one VLAN in use, although we do seem to have left over some VLAN's left showing which are showing up in spanning tree detail I guess because of pvst mode?

1    default                          active    Fa0/2, Fa0/3, Fa0/4, Fa0/5

                                                Fa0/6, Fa0/7, Fa0/8, Fa0/9

                                                Fa0/10, Fa0/11, Fa0/12, Fa0/13

                                                Fa0/14, Fa0/15, Fa0/16, Fa0/17

                                                Fa0/18, Fa0/19, Fa0/20, Fa0/21

                                                Fa0/22, Fa0/23, Fa0/24, Fa0/26

16   VLAN0016                         active

17   VLAN0017                         active

18   VLAN0018                         active

888  Native_Vlan                       active

I've looked at the STP configuration we have and it looks ok from what I've read, but I'm no expert! Can you see anything wrong?                   

[Daran Blackwell]

Anyone? I guess it may be hard to tell from the info I have given

[JohnTylerPearce]

Normally, if you have a spanning tree issue, you will see "MAC Address Flaps" and or "STP Topology Changes".

Do a 'show int trunk', and if you look at 'show span int x/x' and you should be able to see the topology changes on the interfaces, such as BPDUs received. With standard PVST+, basically 802.1D, The root switch will send our BPDUs, and the RPs, on all non-root switches will receive his BPDUs, and push them out their Designated Ports. Unlke RSTP, which allows each and every switch to send out hellos every 2 seconds (kind of like a hello in a dynamic routing protocol), this doesn't happend with 802.1D (PVST+).

Since these ports in question, seem to be user ports, I would like into the computer configuration, Network cable etc. You can also do some testing as well, such as plug a laptop in to one of the port sand see if yours does the same thing etc.

I hope that helped some. Please feel free to ask more questions.

[Daran Blackwell]

Cheers for the reply John - I'll take a look at that cmd ouput.

[rfalconer.sffcu]

Out of curiosity, where the the spanning tree root?

[paul driver]

Hello Daran,

Is portfast enabled on these ports, if not it would be best to enable it , without this feature every time a tcn occurs on the network edge ports will be affected until stp converges.

res

Paul

Please don't forget to rate any posts that have been helpful.

Thanks.