[c-nsp] MPLS Label stack depth and MTU

[Christopher Hunt]

ah, but alas, with 7205 FA-TX you cannot set the IP MTU>1500, so I have 
set the mpls mtu 1508.  The only way to get the extra bits attached is 
to do an end-to-end ping _within_ the vrf, which cannot ping the middle 
hops.  However, someone has suggested that my math is wrong and that in 
fact the L2 header and CRC is 18 bytes, then i need to add an additional 
4 bytes for the VLAN tag which does answer my question.  Problem 
solved!  Now to convince my Service Provider to raise the MTU :-P

Christopher Hunt
ReachONE Internet, Inc.
(888)820-7559



Gert Doering wrote:
> Hi,
>
> On Tue, Apr 29, 2008 at 09:11:15AM -0700, Christopher Hunt wrote:
>   
>> I should have specified that links b,c and d are provided by local fiber 
>> providers (PUDs, telcos, etc.) and are not under our direct control.  
>> They are adding the VLAN tags.
>>     
>
> Underlying transport headers are not relevant for MPLS MTU considerations
> (in IOS, I seem to remember that some versions of JunOS *do* take them into
> account).
>
> Most likely "something in the path" is not really big-MTU compliant - 
> you'll need to ping hop-by-hop and PE-PE (to start the search) with
> df-bit set, as in
>
>   cisco# ping mynexthop size 1526 df-bit
>
> (and this requires "ip mtu" to be the same as "mpls mtu" on the link,
> otherwise the result won't say anthing about MPLS MTU)
>
> gert
>
>