cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
3170
Views
0
Helpful
6
Replies

RSTP strange behaviour

Hello everybody!

I have the following network:

quest1_stp.jpg

I want to make uplinks from 4507 and 3750 always up and Po1 (between 6506 and 4506) as a backup link for vlans 100-4094.

I made:

1) 6506 primary root for vlans 1-2199 and secondary root for 2200-4094

2) 4506 primary root for vlans 2200-4094 and secondary root for 1-2199

3) On the Po1 interface of both 6506 and 4506 i made spanning-tree vlan 100-4094 cost 100

So when I checked this configuration:

6506#sh spanning-tree vlan 109

VLAN0109
  Spanning tree enabled protocol rstp
  Root ID    Priority    24685
             Address     001a.303d.aa00
             This bridge is the root
             Hello Time   2 sec  Max Age 20 sec  Forward Delay 15 sec

  Bridge ID  Priority    24685  (priority 24576 sys-id-ext 109)
             Address     001a.303d.aa00
             Hello Time   2 sec  Max Age 20 sec  Forward Delay 15 sec
             Aging Time 300

Interface        Role Sts Cost      Prio.Nbr Type
---------------- ---- --- --------- -------- --------------------------------
Po1              Desg FWD 3         128.1665 P2p
Po11             Desg FWD 3         128.1672 P2p
Po12             Desg FWD 3         128.1673 P2p

4506#sh spanning-tree vlan 109

VLAN0109
  Spanning tree enabled protocol rstp
  Root ID    Priority    24685
             Address     001a.303d.aa00
             Cost        6
             Port        661 (Port-channel21)
             Hello Time   2 sec  Max Age 20 sec  Forward Delay 15 sec

  Bridge ID  Priority    28781  (priority 28672 sys-id-ext 109)
             Address     000d.bdaa.0680
             Hello Time   2 sec  Max Age 20 sec  Forward Delay 15 sec
             Aging Time 300

Interface           Role Sts Cost      Prio.Nbr Type
------------------- ---- --- --------- -------- --------------------------------
Po1                 Altn BLK 100       128.641  P2p
Po21                Root FWD 3         128.661  P2p
Po22                Desg FWD 3         128.662  P2p

4507#sh spanning-tree vlan 109

VLAN0109
  Spanning tree enabled protocol rstp
  Root ID    Priority    24685
             Address     001a.303d.aa00
             Cost        3
             Port        641 (Port-channel1)
             Hello Time   2 sec  Max Age 20 sec  Forward Delay 15 sec

  Bridge ID  Priority    32877  (priority 32768 sys-id-ext 109)
             Address     c47d.4fc9.2a80
             Hello Time   2 sec  Max Age 20 sec  Forward Delay 15 sec
             Aging Time 300

Interface           Role Sts Cost      Prio.Nbr Type
------------------- ---- --- --------- -------- --------------------------------
Po1                 Root FWD 3         128.641  P2p
Po2                 Desg FWD 3         128.642  P2p

It seems like all is OK, but when I tried to do some spanning tree logging I get this on 4506:

Dec 14 20:49:01.811 MSK: %SPANTREE-5-ROOTCHANGE: Root Changed for vlan 109: New Root Port is Port-channel1. New Root Mac Address is 001a.303d.aa00
Dec 14 20:49:01.823 MSK: %SPANTREE-5-TOPOTRAP: Topology Change Trap for vlan 109
Dec 14 20:49:03.895 MSK: %SPANTREE-5-TOPOTRAP: Topology Change Trap for vlan 109
Dec 14 20:49:03.895 MSK: %SPANTREE-5-ROOTCHANGE: Root Changed for vlan 109: New Root Port is Port-channel21. New Root Mac Address is 001a.303d.aa00
Dec 14 21:13:18.269 MSK: %SPANTREE-5-ROOTCHANGE: Root Changed for vlan 109: New Root Port is Port-channel1. New Root Mac Address is 001a.303d.aa00
Dec 14 21:13:18.277 MSK: %SPANTREE-5-TOPOTRAP: Topology Change Trap for vlan 109
Dec 14 21:19:35.534 MSK: %SPANTREE-5-ROOTCHANGE: Root Changed for vlan 109: New Root Port is Port-channel1. New Root Mac Address is 001a.303d.aa00

Vlan 109 belongs to 4507 and does not generate any topology changes:

4507#sh spanning-tree int fa6/10 det
Port 330 (FastEthernet6/10) of VLAN0109 is designated forwarding
   Port path cost 19, Port priority 128, Port Identifier 128.330.
   Designated root has priority 24685, address 001a.303d.aa00
   Designated bridge has priority 32877, address c47d.4fc9.2a80
   Designated port id is 128.330, designated path cost 3
   Timers: message age 0, forward delay 0, hold 0
   Number of transitions to forwarding state: 1
   The port is in the portfast mode
   Link type is point-to-point by default
   Bpdu filter is enabled
   BPDU: sent 0, received 0

There are no interface flaps in the network also. So I wonder why does it flap between po1 and po21 especially if I look at "New Root Mac Address is 001a.303d.aa00" which is 6506 mac address.

6 Replies 6

Peter Paluch
Cisco Employee
Cisco Employee

Alexander,

You are dealing with a very interesting issue.

My first impression is that the 4506 is cycling between different root ports. The root bridge remains the same but for some reason, the 4506 appears to constantly change its root port.

After reviewing the configuration snippets you have provided, I do not see any obvious problem. There is, however, a thing that caught my attention: the 4507 is configured with the spanning-tree portfast bpdufilter default command. Although that one should not absolutely have any influence on your network, let's keep in mind that this can be a bug. Can you temporarily remove this command and see if the 4506 becomes stable? Removing this command will not modify any RSTP behavior, it will just allow BPDUs to be sent out all ports, including edge ports, which is not a bad thing at all.

What does the same logging reveal on the 4507?

Best regards,

Peter

Peter,

thanks for taking time to answer.

"spanning-tree portfast bpdufilter default" was configured on the 4507 not 4506. Anyway I removed this command from all switches. But nothing changed. It is interesting to note that these rootportchanges doesn't have any fixed time intervals. For example, during the working day, mostly it was OK. But now (evening) it happens every minute or faster (see logs.txt).

show_commands.txt - some output from show commands during such a rootportchange log message.

Best regards,

Alexander.

letsgomets
Level 1
Level 1

Why use the 3rd link on the 4506 if you are costing higher than the other two links?

Why not just eliminate it and allow the root to be learned via 4507 as primary and 3750 as secondary?

letsgomets
Level 1
Level 1

I just answered my own question. I misread the config. You are excluding vlans 1 - 99 from the cost.

It's interesting to note that only 4506 and 4507 log these messages and 3750 and 6506 does not:

10.10.255.2: Dec 20 09:15:35.936 MSK: %SPANTREE-5-TOPOTRAP: Topology Change Trap for vlan 2303
10.10.255.2: Dec 20 09:15:35.936 MSK: %SPANTREE-5-ROOTCHANGE: Root Changed for vlan 2303: New Root Port is Port-channel2. New Root Mac Address is 000d.bdaa.0680

But all the devices catch this topology change in show command:

4507-ORD-L1#sh spanning-tree vlan 2300 det | inc (last change|from)
  Number of topology changes 7 last change occurred 00:36:12 ago
          from Port-channel2

6506-CORE-1#sh spanning-tree vlan 2300 det | inc (last change|from)
  Number of topology changes 615 last change occurred 00:36:18 ago
          from Port-channel12

4506-CORE-2#sh spanning-tree vlan 2300 det | inc (last change|from)
  Number of topology changes 15 last change occurred 00:36:27 ago
          from Port-channel21

3750-KIT-L1#sh spanning-tree vlan 2300 det | inc (last change|from)
  Number of topology changes 555 last change occurred 00:36:33 ago
          from Port-channel2

Here is an update.

This is an ouput from 4507:

Dec 22 22:36:54.044 MSK: STP: VLAN2303 rx BPDU: config protocol = rstp, packet from Port-channel2  , linktype SSTP , enctype 3, encsize 22
Dec 22 22:36:54.048 MSK: STP: VLAN2314 rx BPDU: config protocol = rstp, packet from Port-channel2  , linktype SSTP , enctype 3, encsize 22
Dec 22 22:36:56.048 MSK: STP: VLAN2314 rx BPDU: config protocol = rstp, packet from Port-channel2  , linktype SSTP , enctype 3, encsize 22
Dec 22 22:36:58.048 MSK: STP: VLAN2314 rx BPDU: config protocol = rstp, packet from Port-channel2  , linktype SSTP , enctype 3, encsize 22
Dec 22 22:37:00.044 MSK: STP: VLAN2303 rx BPDU: config protocol = rstp, packet from Port-channel2  , linktype SSTP , enctype 3, encsize 22

And from the root of the spanning tree of the vlans 2303 and 2314 - 4506 all the BPDUs are sent successfully.

So it seems like bpdus are lost on the link. But I wonder how it could so selectively lost vlan2303 bpdus and NOT lost 2314.

And another thing:

4507-ORD-L1#sh int po2 | i (packets input|packets output)
     4901097 packets input, 3694545143 bytes, 0 no buffer
     5105808 packets output, 4287453475 bytes, 0 underruns

4506-CORE-2#sh int po21 | inc (packets input|packets output)
     5104887 packets input, 4286827025 bytes, 0 no buffer
     4900151 packets output, 3692884933 bytes, 0 underruns

No losses, errors, drops or flaps of the interfaces. And packets inputs and outputs are almost equel.