[a-nsp] 7050SX MPLS ?

[Bryan Holloway]

That would be great, Patrik ... I must admit I don't have much 
experience with Martini.

And you make a good point re: the hypervisor ... it's very well possible 
that my IS-IS issue is a red-herring in this instance.


On 1/27/21 3:56 PM, Patrik Olsson wrote:
> Hello,
> 
> depending on your hypervisor, the virtual networ might not be able to support higher MTUs.
> 
> THAT said, in my vEOS lab (KVM with no changed MTUs anywhere), I have runend both MPLS L3VPN PE services as well as martini circuits.
> 
> Want me to share configs?
> 
> Thx
> Patrik
> 
>> On 27 Jan 2021, at 15:55, Bryan Holloway <bryan at shout.net> wrote:
>>
>>
>>
>> On 1/26/21 7:35 PM, Mark Tinka wrote:
>>> On 1/26/21 19:03, Bryan Holloway wrote:
>>>>
>>>> Yeah, I've been bitten by that before ...
>>>>
>>>> While the SVI /30 is IP MTU 1500, the L2 MTU (for tagging, MPLS, etc.) is 9200.
>>> Are you able to either:
>>>      - Raise the MTU on the SVI, or
>>>      - Run IP/MPLS on a/the physical port?
>>
>> So to be clear, I'm currently lab'ing this up on a vEOS instance running 4.25.0.
>>
>> I originally had the IP/MPLS on the "physical" et1 and ran into the issues I saw.
>>
>> Tried changing the L3 MTU on the physical interface to 1504 ... no change in behavior.
>>
>> (Side note: I tried changing it to 1508, and then IS-IS dropped, which I find very peculiar, because IS-IS, being a L2 protocol, shouldn't be affected by the MTU of a L3 interface.)
>>
>> Next I tried changing et1 to a switchport and moving the L3 stuff to an SVI and ran into the same issues. I was at least able to confirm that et1 had an L2 MTU of 9214, and the SVI had an L3 MTU of 1500. Tried changing that to 1504 ... same behavior.
>>
>> (Side note #2: I see the same behavior when changing the L3 MTU to 1508 ... IS-IS drops. Very strange.)
>>
>> Dehr?
>