Re: [c-nsp] MTU issue on a GRE tunnel
Hi, mss should be mtu - 20 - 20 (IP and TCP headers). So try using "ip tcp adjust-mss 1436". Best regards, Jan On 5. 5. 2010 1:18, ML wrote: > I setup a GRE tunnel between two locations served by a Comcast business > class cable modem. > > Config example. The remote end is the same. > > Tunnel73 > ip address yy.yy.yy.yy 255.255.255.252 > ip mtu 1476 > ip tcp adjust-mss 1460 > tunnel source x > tunnel destination z > tunnel path-mtu-discovery > > > The two tunnel endpoints are ME3400s. I expected that this > configuration would reduce the quantity of fragmentations needed by the > router but looks like "IP Input" was up to around 20%. The user > experience was bad since most web pages would fail to load. I thought > "ip tcp adjust-mss" would signal help signal the users OSs to act > appropriately given the situation. > > Have I missed something obvious? > > ___ > cisco-nsp mailing list cisco-nsp@puck.nether.net > https://puck.nether.net/mailman/listinfo/cisco-nsp > archive at http://puck.nether.net/pipermail/cisco-nsp/ signature.asc Description: OpenPGP digital signature ___ cisco-nsp mailing list cisco-nsp@puck.nether.net https://puck.nether.net/mailman/listinfo/cisco-nsp archive at http://puck.nether.net/pipermail/cisco-nsp/
Re: [c-nsp] MTU issue on a GRE tunnel
Config example. The remote end is the same. Tunnel73 ip address yy.yy.yy.yy 255.255.255.252 ip mtu 1476 ip tcp adjust-mss 1460 tunnel source x tunnel destination z tunnel path-mtu-discovery The two tunnel endpoints are ME3400s. I expected that this configuration would reduce the quantity of fragmentations needed by the router but looks like "IP Input" was up to around 20%. The user experience was bad since most web pages would fail to load. I thought "ip tcp adjust-mss" would signal help signal the users OSs to act appropriately given the situation. do fragments increase in "show ip traffic | in frag"? In any case, since you use ip mtu 1476, you should set the TCP MSS at 40 bytes lower (max 1436 in your case) Regards, John ___ cisco-nsp mailing list cisco-nsp@puck.nether.net https://puck.nether.net/mailman/listinfo/cisco-nsp archive at http://puck.nether.net/pipermail/cisco-nsp/
Re: [c-nsp] MTU issue on a GRE tunnel
>-Original Message- >From: cisco-nsp-boun...@puck.nether.net [mailto:cisco-nsp-boun...@puck.nether.net] On Behalf Of ML >The two tunnel endpoints are ME3400s. I believe GRE Tunnels are not supported on ME3400. The packets hit the CPU which is not very fast on that platform. I know 100% for a fact this is the case on 3550s, and AFAIK it applies to all lower-end fix-configuration switches. Ryan Werber ___ cisco-nsp mailing list cisco-nsp@puck.nether.net https://puck.nether.net/mailman/listinfo/cisco-nsp archive at http://puck.nether.net/pipermail/cisco-nsp/
[c-nsp] MTU issue on a GRE tunnel
I setup a GRE tunnel between two locations served by a Comcast business class cable modem. Config example. The remote end is the same. Tunnel73 ip address yy.yy.yy.yy 255.255.255.252 ip mtu 1476 ip tcp adjust-mss 1460 tunnel source x tunnel destination z tunnel path-mtu-discovery The two tunnel endpoints are ME3400s. I expected that this configuration would reduce the quantity of fragmentations needed by the router but looks like "IP Input" was up to around 20%. The user experience was bad since most web pages would fail to load. I thought "ip tcp adjust-mss" would signal help signal the users OSs to act appropriately given the situation. Have I missed something obvious? ___ cisco-nsp mailing list cisco-nsp@puck.nether.net https://puck.nether.net/mailman/listinfo/cisco-nsp archive at http://puck.nether.net/pipermail/cisco-nsp/
