RSTP and long time-outs

Question

Hi,

I'm using two 6024's connected by using a LAG (port 23-24). Four 5324's are connected with port 1 to the first 6024 and port 2 to the second 6024. I enabled RSTP at each switch and configured the switch ports involved as trunks for every available VLAN.

Everything works fine but when I for instance disable port 1 at the 5324 is takes almost 30 seconds to detect that there is a redundant path to the 6024 resulting in time-outs. When enabling the port again it again takes about 30 seconds before traffic is allowed. Can this downtime of the link be reduced?

Port/RSTP config 6024's:

spanning-tree mode rstp
spanning-tree pathcost method short
interface ethernet g3
spanning-tree cost 4
exit
interface port-channel 1
switchport mode trunk
exit
interface range ethernet g(1-4)
switchport mode trunk
exit
vlan database
vlan 2,4-6,9-10,200
exit
interface range ethernet g(1-4)
switchport trunk allowed vlan add 2
exit
interface range port-channel 1
switchport trunk allowed vlan add 2
exit
interface range ethernet g(9-22)
switchport access vlan 4
exit
interface range ethernet g(1-4)
switchport trunk allowed vlan add 4
exit
interface range port-channel 1
switchport trunk allowed vlan add 4
exit
interface range ethernet g(1-4)
switchport trunk allowed vlan add 5
exit
interface range port-channel 1
switchport trunk allowed vlan add 5
exit
interface range ethernet g(1-4)
switchport trunk allowed vlan add 6
exit
interface range port-channel 1
switchport trunk allowed vlan add 6
exit
interface range ethernet g(1-4)
switchport trunk allowed vlan add 9
exit
interface range port-channel 1
switchport trunk allowed vlan add 9
exit
interface range ethernet g(1-4)
switchport trunk allowed vlan add 10
exit
interface range port-channel 1
switchport trunk allowed vlan add 10
exit
interface range ethernet g(5-8)
switchport access vlan 200
exit
interface range ethernet g(1-4)
switchport trunk allowed vlan add 200
exit
interface range port-channel 1
switchport trunk allowed vlan add 200
exit
interface range ethernet g(23-24)
channel-group 1 mode auto
exit

Port/RSTP config 5324's:

spanning-tree mode rstp
spanning-tree bpdu filtering
interface ethernet g1
speed 1000
duplex full
exit
interface ethernet g1
switchport mode trunk
exit
interface ethernet g2
switchport mode trunk
exit
vlan database
vlan 2,4-6,9-10,200
exit
interface range ethernet g(3-24)
switchport access vlan 2
exit
interface range ethernet g(1-2)
switchport trunk allowed vlan add 2
exit
interface range ethernet g(1-2)
switchport trunk allowed vlan add 4
exit
interface range ethernet g(1-2)
switchport trunk allowed vlan add 5
exit
interface range ethernet g(1-2)
switchport trunk allowed vlan add 6
exit
interface range ethernet g(1-2)
switchport trunk allowed vlan add 9
exit
interface range ethernet g(1-2)
switchport trunk allowed vlan add 10
exit
interface range ethernet g(1-2)
switchport trunk allowed vlan add 200

Thanks,

Hans

hansvz · Answer

Hi Adam,   I double checked all switches and they were correctly configured for RSTP. Then I decided to save the running-config at all the switches and gave a 'reload' command at each of them, starting with the 6024's.   Guess what? It is working flawlessly! Within .5 to 2 seconds there is a link again by using the alternate path when the primary path is unavailable.   I did not set a root switch manually, will this improve the reconvergence time any further? I thought setting a root switch was only applicable to MSTP and not RSTP?   Regards,   Hans

Adam N · Answer

Hi Hans,   It should not take more than a few seconds to reconverge.  I would check all of your switch configs and ensure they are all enabled for RSTP.  It would only take 1 switch to send an 802.1D BPDU and the other switches on that segment would also revert back to transmitting 802.1d bpdu's which would explain the 30 second delay.   Also looking at the configs below you have not manually set a root switch, I would suggest you set one of your PC6024 switches to be root so that the switches do not have to elect a root themselves.   Hope this helps   Regards

Adam N · Answer

Hi Hans,

It is important to manually set which switch you would like to be the root switch in all spanning-tree environments. By default each switch has a BID (Bridge ID) which is default priority of 32,768 + the switch MAC address. If left at default the switch with lowest BID will be elected as the root switch. Now as you can see this is not the best way of electing the most important switch on your network.

The root switch should be the switch closest to the centre of your spanning-tree topology, looking at your topology either PC6024 switch would be a good choice.

When you powered off / on each switch, the order in which you brought them back up also decided on which switch was selected as root. So as you can see it is important to manually set the root.

On the switch you would always like to be root, type the following:-

spanning-tree priority 4096

The priority you use can be anything as long as its the lowest priority on your network.

Once you set this it should work fine everytime.

Thanks

hansvz · Answer

Hi Adam,   The first 6024 has now been configured with an spanning-tree priority of 4096 and the second 6024 with a priority of 8192. All 5324's still have the default priority. It is working great, even if I turn off the 6024 with the lowest priority reconvergence is fast.   Thanks for helping me out!   Hans

Networking General

RSTP and long time-outs

Was this post helpful?