Spanning Tree failback question/issue

[Trooper]

I am seeing about a 15 second delay passing traffic when Meraki RSTP fails back.

I would expect it to be much faster.

I have 3 switches setup to form a triangle, two Cisco catalysts and one Meraki.

All switches are running RSTP.

One of the Cisco’s is the Primary Root bridge, the other Cisco is the Secondary Root bridge.

My test machine is connected to the Meraki, and I have a server connected to the Cisco Root.

I run a constant ping from the test machine on the Meraki to the server connected to the Cisco Root.

If I shut down the designated port on the Cisco Root facing the Meraki, traffic fails over fairly quickly to the backup link. Maybe within 1 second.

When I no shut the port, traffic stops passing for about 15 seconds.

Is this expected?

Why is there such a long delay?

The Meraki logs show the port switch to root about 4 seconds after the port is up.

But it takes just over 15 seconds for traffic to start passing.

As a comparison, I used another Catalyst switch running RSTP in place of the Meraki, and the failback is almost immediate.

[JacekJ]

First things first, what Catalysts are you using?

Are you sure you are running RSTP?

[Trooper]

This is a lab environment. Below is the info.

Yes, everything is running RSTP.

Core-Pri is a Cisco WS-C3560-8PC, running version 15.0(2)SE11

Core-Sec is a Cisco WS-C3560-24TS, running version 12.2(55)SE11

The Meraki is an MS-220-8P, running version 14.33

Here is the SPT config from the Cisco's:

Core-Pri#sh run | in spann
spanning-tree mode rapid-pvst
spanning-tree extend system-id
spanning-tree vlan 1-4094 priority 4096

Core-Pri#sh spann vlan 1

VLAN0001
Spanning tree enabled protocol rstp
Root ID Priority 4097
Address c062.6b13.9a00
This bridge is the root
Hello Time 2 sec Max Age 20 sec Forward Delay 15 sec

Core-Sec#sh run | in spann
spanning-tree mode rapid-pvst
spanning-tree extend system-id
spanning-tree vlan 1-4094 priority 8192

Core-Sec#sh spann vlan 1

VLAN0001
Spanning tree enabled protocol rstp
Root ID Priority 4097
Address c062.6b13.9a00
Cost 19
Port 26 (FastEthernet0/24)
Hello Time 2 sec Max Age 20 sec Forward Delay 15 sec

[JacekJ]

OK, so you are using Rapid-PVST, which is not RSTP that is used by Meraki.

Please read the below documentation, are you blocking VLAN1 anywhere?

https://documentation.meraki.com/MS/Deployment_Guides/Advanced_MS_Setup_Guide#Rapid-PVST

[Trooper]

Thanks for the info.

I understand that Meraki and Cisco use different versions of Rapid Spanning Tree.

I am not blocking VLAN1 anywhere.  The Meraki does fail back so it is working. I am wondering why there is that delay. The Meraki logs show the port transitioning to Root much faster than the time it takes for it to start passing traffic. With the 3rd Cisco in place of the Meraki, the transition is not even noticeable. Does the Cisco continue to pass traffic on the old Root port while it is learning the new Root port, where Meraki stops traffic until it learns (and then some)?   Maybe that is the question I should be asking. The docs I have found seem to mention failover, but not so much for failback scenarios.

[PhilipDAth · Accepted solution]

On the Catalysts change to mst.

spanning-tree mode mst

[Trooper]

Very good.  That makes the failback happen quick. 

Good tip. 

I am doing this in a lab, in prep for a larger production rollout.

So I'll have to think this through a little more as the real environment is much larger than the lab.

But that definitely made a difference.

Good advice.

[RaphaelL]

@PhilipDAth , I'm trying to move from RSTP ( spanning-tree mode rapid-pvst ) to MSTP single instance as suggested. What are the benefits ? The convergence time should be the same ( 3 hello x 2 seconds ) , no ? Am I missing something obvious ?

Ps : I have read your posts on cisco also , but I can't make my mind on it 🙂 

Thanks , 

[PhilipDAth]

A reduction in spanning tree outages in a mixed Cisco Enterprise and Meraki environment.  If you don't mind intermittent outages then you can leave it as RSTP.