Re: [c-nsp] q-in-q mtu

2010-10-21 Thread Tony
--- On Fri, 22/10/10, Chris Wopat wrote: > > On Thu, 21 Oct 2010, Daniel > Hooper wrote: > > > > The 7206 doesn't accept anything higher than 1500 on > that interface. > > Some 12.2S trains allow you to up the MTU on 7200 to 1530. > I'm > currently running SB, I'm unsure if SR* can do it or not.

Re: [c-nsp] q-in-q mtu

2010-10-21 Thread Antonio Querubin
On Thu, 21 Oct 2010, Daniel Lacey wrote: Isn't 1490 the "magic" MTU for PPPoE? Perhaps. But this is supposed to be a layer 2 connection via ethernet from us to the customer. PPPoE isn't in use. Just some switches and q-in-q in between. Antonio Querubin 808-545-5282 x3003 e-mail/xmpp: t

Re: [c-nsp] q-in-q mtu

2010-10-21 Thread Daniel Lacey
Isn't 1490 the "magic" MTU for PPPoE? |--- | Dan Lacey daniel_p_la...@yahoo.com | PGP Key: 0xFE94668F @ http://pgp.mit.edu or http://keyserver.pgp.com | PGP Key fingerprint: 8A97 2996 266D A21C 0277 54EF 40D5 2B80 FE94 668F |

Re: [c-nsp] q-in-q mtu

2010-10-21 Thread Antonio Querubin
On Thu, 21 Oct 2010, David Freedman wrote: Problem is the DEC chip found on PA-FE-TX , it has a hard limit of 1530 It's actually the built-in FastEthernet port instead of a port adapter: #show controllers fastEthernet 0/0 Interface FastEthernet0/0 Hardware is DEC21140A PDU (i.e on the wire,

[c-nsp] q-in-q mtu

2010-10-21 Thread Chris Wopat
> On Thu, 21 Oct 2010, Daniel Hooper wrote: > > The 7206 doesn't accept anything higher than 1500 on that interface. Some 12.2S trains allow you to up the MTU on 7200 to 1530. I'm currently running SB, I'm unsure if SR* can do it or not. http://puck.nether.net/pipermail/cisco-nsp/2007-March/03903

Re: [c-nsp] q-in-q mtu

2010-10-21 Thread David Freedman
Problem is the DEC chip found on PA-FE-TX , it has a hard limit of 1530 PDU (i.e on the wire, with headers), IOS prevented you changing the mtu on the interface (which is usually the SDU, i.e without the ethernet frame overhead) because it was so close to the edge, but with advent of "tag-switching

Re: [c-nsp] q-in-q mtu

2010-10-21 Thread Christophe Lucas
Le 20/10/2010 19:42, Antonio Querubin a écrit : A new carrier will be handing us customer connections as q-in-q vlans. However, during our initial network validation we noticed what seem like possible MTU issues. Pings work fine but HTTP connections to various sites is flakey - more so for IPv4

Re: [c-nsp] q-in-q mtu

2010-10-21 Thread Antonio Querubin
On Thu, 21 Oct 2010, Christophe Lucas wrote: You could be able to set mtu higher than 1500, so that you can handle q-in-q frames. 1500 is the highest MTU the router will accept for configuration on its FastEthernet interfaces. It doesn't have any GigabitEthernet interfaces. I am interest

Re: [c-nsp] q-in-q mtu

2010-10-21 Thread Christophe Lucas
Le 21/10/2010 11:21, Antonio Querubin a écrit : On Thu, 21 Oct 2010, Daniel Hooper wrote: I was under the impression that interfaces handling Q-in-Q needed atleast an MTU of 1504 bytes. The 7206 doesn't accept anything higher than 1500 on that interface. I did some testing with the carrier t

Re: [c-nsp] q-in-q mtu

2010-10-21 Thread Antonio Querubin
On Thu, 21 Oct 2010, Daniel Hooper wrote: I was under the impression that interfaces handling Q-in-Q needed atleast an MTU of 1504 bytes. The 7206 doesn't accept anything higher than 1500 on that interface. I did some testing with the carrier today and we found that 1490 was the highest MTU

[c-nsp] q-in-q mtu

2010-10-20 Thread Antonio Querubin
A new carrier will be handing us customer connections as q-in-q vlans. However, during our initial network validation we noticed what seem like possible MTU issues. Pings work fine but HTTP connections to various sites is flakey - more so for IPv4 than IPv6 oddly enough. The customer circuit