spanning tree problem

[paul winterbourne]

Hi all,

I am new here and I am hoping one of you experts can help me out... I have implemented a new gigabuit backbone into our seven switch backbone loop. I am basically replacing old intel 510's with a couple of netgears and hp procurve 2524's. However I am having awful awful problems getting stp to settle down. I have set the bridge prioirity lowest on a netgear gsm7224 switch and all the other switches accept it (including the procurves). I am finding though that stp is reconverging about every half an hour even though i can't find any broken links in my backbone. Not only that but underneath one of the backbone procurves is one of the intels plugged in using ethernet (port 24 to port 24). I set port 24 on the procurve to be an edge port however I am still receiving bpdu ignored messages on the intel all the time. Does anyone know of any solutions|?

many thanks,

Paul

[DaGuru]

Hi Paul,

Can you please attach a network drawing that illustrates your physcial network topology? Please include STP priorities for each case where the default values have been changed. This will help us to visualize your setup.

Also, configuring the ProCurve for edge-mode only puts the port into a forwarding state. It does not stop BPDU propegation.

Thanks,
Dennis

[paul winterbourne]

Hi Dennis,

Thanks for your reply, I am a bit stumped here and it is a bit of a nightmare as this is the first big network project I have been leading. I have included the diagrams of our network (please find attached). Start with the backbone config and the other two diagrams are of the switches which have more than one switch at each area. I have disabled spanning tree on all switches apart from the back bone switches. The last intel switch in the network is in the main office and runs 802.1d. All the other backbone switches are set to run rstp but are therefore running 802.1d for compatibility purposes. I have not set any edge ports apart from port 24 hp switch at poultry (which is linked to an intel 510 100mbps 24 port switch which is coming up with bpdu ignored message every 2 seconds in the event log). As for stp settings they are as follows:

AlL backbone hp 2524â s are set to have rstp on

Cost = 10 and priority = 128 this is set on all HP 2524 backbone fibre ports apart from the lab where one fibre port is set to have a cost of 64000 (this was so that I was basically stp to force this port as being blocked when stp kicks in)

The two netgears are as follows (on all fibre ports):

Path cost = 2 and priority = 128

I have set the netgear in the c site to have a bridge priority of 8196 so that it is the forced root switch.

I have disconnected between the abattoir and the roofspace because e were having loss of network problems and I though breaking the spanning tree would fix it. However it turned out that on one of the poultry switches (not the backbone one) the uplink port was set to 10 mb half duplex and had all our production network traffic going through it so I suspect that the problem was more due to this than stp.

I am at a loss to find out how to stop bpduâ s forwarding on certain ports, can this be done?

Thanks for all your help, itâ s been a real steep learning curve and sleepless nights. Its really good to talk to someone who can answer these sort of questions. Could you also help me in understanding stp in terms of how many switches you need to enable it on? As I said above I have enabled it on all backbone switches (i.e. 7 switches for one big loop). DO I actually need it on all seven switches? or could I just set it on 2 switches? I do realise though that apparently in the netgear firmware I am using if I turn stp off they won't forward on bpdu's (this is fixed in later firmware) so this is obviously theoretical in my case at the moment

Paul

[paul winterbourne]

another attachment

[paul winterbourne]

another attachment for roofspace

[DaGuru]

Hi Paul,

I am sorry for not getting back to you sooner, but I am currently working with a customer while onsite in China. Therefore, here are my fragmented thoughts on your issues. If you would like further explenation on anything I cover or may have missed, please let me know.

#1. I agree that the missconfiguration of port duplex would have possibly caused spanning tree to continually reconverge.

#2. From what I have read on the 2500 series switch, it lacks the ability to block BPDUs on specific ports. I am not familiar with the Intel switch, but I would suggest disabling STP on the Intel. This may prevent further propagation of BPDUs from that point further down the chain.

#3. I would suggest reconnecting the uplinks and let spanning tree do its thing. If you see any reconvergence, start by checking to see if your root switch adopts a new MAC address. Sometimes you can have competing roots, but I suspect that this is not the case because it only happens about once every ½ hour.

#4. Any switch that does not participate in the tree (any switch not directly in the loop) does not need to be running spanning tree.

#5. Depending upon how your physical cabling is run, you might consider building out your backbone by choosing a primary and secondary root, then link them into the rest of your backbone so that you form a logical pyramid of sorts.

For more information on spanning tree... Take a look at this document:
http://www.hp.com/rnd/training/technical/MSTP.htm?jumpid=reg_R1002_USEN