Loss of link on Procurve 5406zl

[digone52]

Hi. I am a bit of a novice at diagnosing faults on our network, and I'd really appreciate some assistance.

I am getting "Loss of link" messages from our 5406zl switches, and seeing momentary drop-outs of ports reported from our VMWare servers, which funnily enough are different ports to those mentioned in the event log of the switch.

We have 2 x 5406zl in one building as core switches and 2 x 5308xl in another as core switches, which are all connected tegether in a mesh.

I can see no reason for these loss of link messages, and the lost links come back promptly.

We have 32 Procurve edge switches connected to these core switches.

One think - if I firmware update an edge switch - any edge switch it would appear - I seem to get excessive broadcasts noted on virtually all of the switches, which carms down after 3 minutes. This behaviour can also cause loss of links on  our 5406zl.

I have the use of Procurve Manager 4 +, and have looked at our topology and removed some errors in terms of loops, and it now looks clean - I can't see any issues.

Re: STP, the 5406zl's and 5308xl's have this... "spanning-tree Mesh priority 0" as their only STP line I can see.

We are going down the route of making the edge switches fault tolerant, by (for instance) on a

2510-48 using below...

spanning-tree 1 bpdu-protection

....

spanning-tree 48 bpdu-protection
spanning-tree config-name "uplinks"
spanning-tree config-revision 1
spanning-tree instance 1 vlan 1 2 6
spanning-tree instance 1 51 priority 1
spanning-tree instance 1 52 priority 2

Where ports 51 and 52 are the uplinks going to each 5406zl switches. You get a block on port 52 until port 51 drops, and then 52 becomes live.

Not all switches are in this configuration yet. 

Re: the firmware updating issue mentioned above, this can happen on a switch with the above config, and switches without - it appears to happen on all switches. It might also happen on a reboot of an edge switch. I don't want to reboot switches too often if it causes disruption to our vmware infrastructure.

I would really welcome any comments on how I go about fault finding this, or if you can help me, what further information you'd need.

[Vince_Whirlwind]

Show us the "drop out" events in the switch log.

Also, show us the spanning-tree priorities of each switch in your 4-switch "core".

[Vince_Whirlwind]

Also, I hope for your sake the message you are getting isn't this one:

 Slot I: Msg loss detected - no ack for seq #            39128