Solved

S-Tag over VOSS Fabric using VSP 7400 Switches


Hi All

 

TLDR: We cannot get S-Tag vlans to pass over the Fabric between our two VSP 7400 switches.

 

We have the following deployed in our network;

 

2 x Extreme VSP 7400 Software version 8.0.5.1

These are joined by two 100Gb cables.

ISIS is enabled

ISIS Spbm 1

The Fibre provider port is Flex-uni enabled and Encapsulation Dot1Q

The Router uplink is Flex-uni enabled and Encapsulation Dot1Q

As a side note we tried removed Encapsulation Dot1Q and it made no difference. 

 

Scenario:

 

We have a fibre provider that is handing services over as S-Tags.

Currently over this setup we are able to pass C-Tags but cannot pass S-Tags.

We can learn MAC address entries when using S-Tag on both switches across the I-SID but not on the edge devices these VLANS terminate onto.

 

Configuration we tried to get it working:

 

Switch 1 - Where fibre provider plugs into

Create I-SID 1000 

c-vid 1000 port 1/1 (Fibre provider)

 

Switch 2 - Where the uplink Router plugs into

Create I-SID 1000 

c-vid 1000 port 1/5 (uplink to Router)

 

This does not work unfortunately, if we drop the S-Tag identifier we can learn MAC addresses and push L3 traffic across. 

 

Things we have tried alternative to config changes:

 

Deployed additional 2 x Extreme VSP 7400s and upgraded them to Software version 8.3.0.0

 

Even on this version we cannot get it working.

 

I have followed this thread and downloaded the PDF and according to that we are doing everything correctly:

 

icon

Best answer by Roger Lapuh 5 May 2021, 09:02

QinQ packet coming from the provider, otherwise VSP does not know to look at the q-tag. VSP does not support QinQ natively at this point. This would require a feature request.

Roger

View original

10 replies

Userlevel 5

I personally re-verified the QinQ setup using VSP7400s and it works correctly. From the packet captures sent to GTAC it would appear that the ethertype in use was not 8100 but 88a8.

Userlevel 4

Ashlee, would you mind sending me the GTAC case number to rlapuh@extremenetworks.com so I can sync with the GTAC engineer. Thanks Roger

Obviously I can’t debug your case here. If you are following everything according to Ludovico’s document, it should work, otherwise I suggest to open a GTAC case. This is not the first time an S-TAG is transported over an SPB fabric. 

Thanks Roger

Obviously.

We have a GTAC case opened.

If you are following everything according to Ludovico’s document, it should work.

I followed this to the T same firmware and configuration, it did not work sadly.

 

Will see what GTAC comes back with, thanks for the replies. 

Userlevel 4

Obviously I can’t debug your case here. If you are following everything according to Ludovico’s document, it should work, otherwise I suggest to open a GTAC case. This is not the first time an S-TAG is transported over an SPB fabric. 

Thanks Roger

Okay I see.

Let me throw a curve ball in :) 

We have our own lab setup that uses the following:

 

0x8100 is kept across all devices involved. The edge routers are mikrotiks with service identifier turned on (S-Tag).

 

Edge router A 0x8100 - VLAN 3490

Edge router B 0x8100 - VLAN 3490

 

Extreme Switch 1 

 

i-sid 3490

c-vid 3490 port 1/7 - Edge router A

 

Extreme Switch 2 

 

i-sid 3490

c-vid 3490 port 1/32 - Edge router B

 

We are unable to pass S-Tag across the fabric with this setup.

MAC Table from SW1

************************************************************************************
                Command Execution Time: Wed May 05 10:54:22 2021 SAST
************************************************************************************

========================================================================================================
                                               I-SID Fdb Table
========================================================================================================
I-SID     STATUS  MAC-ADDRESS        INTERFACE        TYPE       DEST-MAC           BVLAN DEST-SYSNAME
--------------------------------------------------------------------------------------------------------
3490      learned 48:8f:5a:64:d1:eb  Port-1/49        REMOTE     00:db:dc:01:00:02  4052  JHB-Core-Switch-2
3490      learned b8:69:f4:6a:1e:44  c3490:1/7        LOCAL      00:00:00:00:00:00  0

c: customer vid   u: untagged-traffic

All 2 out of 2 Total Num of i-sid FDB Entries displayed
 

 

MAC Table from SW2

                Command Execution Time: Wed May 05 10:53:34 2021 GMT
************************************************************************************

========================================================================================================
                                               I-SID Fdb Table
========================================================================================================
I-SID     STATUS  MAC-ADDRESS        INTERFACE        TYPE       DEST-MAC           BVLAN DEST-SYSNAME
--------------------------------------------------------------------------------------------------------
3490      learned 48:8f:5a:64:d1:eb  c3490:1/32       LOCAL      00:00:00:00:00:00  0
3490      learned b8:69:f4:6a:1e:44  Port-1/49        REMOTE     00:db:dc:01:00:01  4052  JHB-Core-Switch-1

c: customer vid   u: untagged-traffic

All 2 out of 2 Total Num of i-sid FDB Entries displayed
 

As you can see we are learning MAC addresses across the fabric between the switches but not from the routers:

 

Edge Router A

Edge router B

 

 

Userlevel 4

yes, then the S-tag will be used to classify the VLANs.

Okay I see, this is a downer as the reason we went with these switches was to pass S-Tags but I understand what you are saying.

I do not know the ethertype the provider is handing over but I assume it is either 0x88a8 or 0x9100.

 

Let me confirm that, if they can accommodate a 0x8100 ethertype then am I correct to assume the fabric will have no issues passing the S-Tag?

Userlevel 4

QinQ packet coming from the provider, otherwise VSP does not know to look at the q-tag. VSP does not support QinQ natively at this point. This would require a feature request.

Roger

Did you make sure the Ethertype of the QinQ packet is 0x8100, so that the VSP can see this packet as a tagged packet? Thanks Roger

The ethertype of the SPBM is 0x8100 if that is what you are referring to.

Or do you mean the packet coming in from the Fibre provider side?

 

Userlevel 4

Did you make sure the Ethertype of the QinQ packet is 0x8100, so that the VSP can see this packet as a tagged packet? Thanks Roger

Reply