Mailing List Archive

MTU challenges on Cisco 7206VXR
We're mostly PPPoA over an OC-3, but we do have some PPPoE over the OC-3,
and about ~150 customers using PPPoE coming in on NPE-400's Ethernet port.
We have a Cisco 7206VXR with 12.2(26).

We've had a few reports about the last two years, as we added PPPoE support,
that large packets don't get through. The answer was always to lower the
MTU to 1400 on their CPE (mostly SonicWall, some WatchGuard) and the problem
was solved.

We have an existing customer moving from PPPoA to PPPoE coming in on the
OC-3with a new consultant and he's pretty sure that "it should just work".
What's odd is that the ~150 PPPoE customers coming in over the Ethernet port
and we haven't had any PPPoE issues. So I'm guessing something may be
configured on the ATM interface that's not an issue on the Ethernet
interface.

Has anyone run into this problem, and if so, is there something I can do on
our BRAS to resolve this?

Regards,

Frank

_______________________________________________
cisco-bba mailing list
cisco-bba@puck.nether.net
https://puck.nether.net/mailman/listinfo/cisco-bba
Re: MTU challenges on Cisco 7206VXR [ In reply to ]
ATM has an MTU of around 4470 buytes, fast ethernet is 1500 tops.
Gigabit ethernet can be upto ~9000 bytes if jumbo frames are enabled.

At the most you should only need to shave 8 bytes off of your MTU for
PPPoE to account for additional PPPoE headers.

ATM, whether you run PPPoA or PPPoEoA, should be less likely to
experience MTU issues. PPPoEoE over gigabit ethernet should not
require any MTU adjustments, but over fast ethernet you should drop
your MTU to 1492 at both ends.

Cheers,
Tom

On 13/03/2008, at 6:43 AM, Frank Bulk - iNAME wrote:

> We're mostly PPPoA over an OC-3, but we do have some PPPoE over the
> OC-3,
> and about ~150 customers using PPPoE coming in on NPE-400's Ethernet
> port.
> We have a Cisco 7206VXR with 12.2(26).
>
> We've had a few reports about the last two years, as we added PPPoE
> support,
> that large packets don't get through. The answer was always to
> lower the
> MTU to 1400 on their CPE (mostly SonicWall, some WatchGuard) and the
> problem
> was solved.
>
> We have an existing customer moving from PPPoA to PPPoE coming in on
> the
> OC-3with a new consultant and he's pretty sure that "it should just
> work".
> What's odd is that the ~150 PPPoE customers coming in over the
> Ethernet port
> and we haven't had any PPPoE issues. So I'm guessing something may be
> configured on the ATM interface that's not an issue on the Ethernet
> interface.
>
> Has anyone run into this problem, and if so, is there something I
> can do on
> our BRAS to resolve this?
>
> Regards,
>
> Frank
>
> _______________________________________________
> cisco-bba mailing list
> cisco-bba@puck.nether.net
> https://puck.nether.net/mailman/listinfo/cisco-bba

_______________________________________________
cisco-bba mailing list
cisco-bba@puck.nether.net
https://puck.nether.net/mailman/listinfo/cisco-bba
Re: MTU challenges on Cisco 7206VXR [ In reply to ]
Ok, I agree. So where are the problems coming from?

I do see that on the "root" ATM interface I have the following entered:
interface ATM4/0
mtu 1504
no ip address
no atm ilmi-keepalive
!
And the PVCs are defined under a sub-interface as follows:
interface ATM4/0.100 multipoint
pvc 0/261
encapsulation aal5mux ppp Virtual-Template1
!
pvc 0/262
encapsulation aal5snap
protocol pppoe
!
pvc 0/263
encapsulation aal5mux ppp Virtual-Template1
!

Is it possible that the PVC connections are inheriting the 1504 for the
ATM4/0 interface and that I should delete that line altogether?

Frank

-----Original Message-----
From: Tom Storey [mailto:tom@snnap.net]
Sent: Wednesday, March 12, 2008 10:01 PM
To: frnkblk@iname.com
Cc: cisco-bba@puck.nether.net
Subject: Re: [cisco-bba] MTU challenges on Cisco 7206VXR

ATM has an MTU of around 4470 buytes, fast ethernet is 1500 tops.
Gigabit ethernet can be upto ~9000 bytes if jumbo frames are enabled.

At the most you should only need to shave 8 bytes off of your MTU for
PPPoE to account for additional PPPoE headers.

ATM, whether you run PPPoA or PPPoEoA, should be less likely to
experience MTU issues. PPPoEoE over gigabit ethernet should not
require any MTU adjustments, but over fast ethernet you should drop
your MTU to 1492 at both ends.

Cheers,
Tom

On 13/03/2008, at 6:43 AM, Frank Bulk - iNAME wrote:

> We're mostly PPPoA over an OC-3, but we do have some PPPoE over the
> OC-3,
> and about ~150 customers using PPPoE coming in on NPE-400's Ethernet
> port.
> We have a Cisco 7206VXR with 12.2(26).
>
> We've had a few reports about the last two years, as we added PPPoE
> support,
> that large packets don't get through. The answer was always to
> lower the
> MTU to 1400 on their CPE (mostly SonicWall, some WatchGuard) and the
> problem
> was solved.
>
> We have an existing customer moving from PPPoA to PPPoE coming in on
> the
> OC-3with a new consultant and he's pretty sure that "it should just
> work".
> What's odd is that the ~150 PPPoE customers coming in over the
> Ethernet port
> and we haven't had any PPPoE issues. So I'm guessing something may be
> configured on the ATM interface that's not an issue on the Ethernet
> interface.
>
> Has anyone run into this problem, and if so, is there something I
> can do on
> our BRAS to resolve this?
>
> Regards,
>
> Frank
>
> _______________________________________________
> cisco-bba mailing list
> cisco-bba@puck.nether.net
> https://puck.nether.net/mailman/listinfo/cisco-bba


_______________________________________________
cisco-bba mailing list
cisco-bba@puck.nether.net
https://puck.nether.net/mailman/listinfo/cisco-bba
Re: MTU challenges on Cisco 7206VXR [ In reply to ]
Frank Bulk - iNAME wrote:
> Ok, I agree. So where are the problems coming from?
>
> I do see that on the "root" ATM interface I have the following entered:
> interface ATM4/0
> mtu 1504
> no ip address
> no atm ilmi-keepalive
> !
> And the PVCs are defined under a sub-interface as follows:
> interface ATM4/0.100 multipoint
> pvc 0/261
> encapsulation aal5mux ppp Virtual-Template1
> !
> pvc 0/262
> encapsulation aal5snap
> protocol pppoe
> !
> pvc 0/263
> encapsulation aal5mux ppp Virtual-Template1
> !
>
> Is it possible that the PVC connections are inheriting the 1504 for the
> ATM4/0 interface and that I should delete that line altogether?


I'd leave the MTU at default on the ATM interface and change the MTU on
your virtual-template to 1492. We have tens of thousands of
DSL/Ethernet subs using PPPoE/oA with no MTU problems at all. Usually
the device terminating the PPPoX is smart enough to know to adjust the
MTU... some not so much.

There are some devices in which you have to clamp the mss because the
client can't figure it out for some reason... however, that should be
few and far between.


--
Robert Blayzor
INOC
rblayzor@inoc.net
http://www.inoc.net/~rblayzor/

> SELECT * FROM users WHERE clue > 0
0 rows returned
_______________________________________________
cisco-bba mailing list
cisco-bba@puck.nether.net
https://puck.nether.net/mailman/listinfo/cisco-bba
Re: MTU challenges on Cisco 7206VXR [ In reply to ]
Thanks for the feedback. So, do you think that 'mtu 1504' is a valid
configuration item and I should leave that alone? I would prefer not to
drop 8 bytes from my PPPoA customers if removing the 'mtu 1504' entry would
solve the problem.

Frank

-----Original Message-----
From: Robert Blayzor [mailto:rblayzor.bulk@inoc.net]
Sent: Thursday, March 13, 2008 7:39 AM
To: frnkblk@iname.com
Cc: cisco-bba@puck.nether.net
Subject: Re: [cisco-bba] MTU challenges on Cisco 7206VXR

Frank Bulk - iNAME wrote:
> Ok, I agree. So where are the problems coming from?
>
> I do see that on the "root" ATM interface I have the following entered:
> interface ATM4/0
> mtu 1504
> no ip address
> no atm ilmi-keepalive
> !
> And the PVCs are defined under a sub-interface as follows:
> interface ATM4/0.100 multipoint
> pvc 0/261
> encapsulation aal5mux ppp Virtual-Template1
> !
> pvc 0/262
> encapsulation aal5snap
> protocol pppoe
> !
> pvc 0/263
> encapsulation aal5mux ppp Virtual-Template1
> !
>
> Is it possible that the PVC connections are inheriting the 1504 for the
> ATM4/0 interface and that I should delete that line altogether?


I'd leave the MTU at default on the ATM interface and change the MTU on
your virtual-template to 1492. We have tens of thousands of
DSL/Ethernet subs using PPPoE/oA with no MTU problems at all. Usually
the device terminating the PPPoX is smart enough to know to adjust the
MTU... some not so much.

There are some devices in which you have to clamp the mss because the
client can't figure it out for some reason... however, that should be
few and far between.


--
Robert Blayzor
INOC
rblayzor@inoc.net
http://www.inoc.net/~rblayzor/

> SELECT * FROM users WHERE clue > 0
0 rows returned

_______________________________________________
cisco-bba mailing list
cisco-bba@puck.nether.net
https://puck.nether.net/mailman/listinfo/cisco-bba
Re: MTU challenges on Cisco 7206VXR [ In reply to ]
Should I be applying
ip tcp adjust-mss 1492
to the Virtual-Template, or
mtu 1492

I have a potential maintenance window tonight and I would like to sneak this
in there.

Frank

-----Original Message-----
From: cisco-bba-bounces@puck.nether.net
[mailto:cisco-bba-bounces@puck.nether.net] On Behalf Of Frank Bulk
Sent: Thursday, March 13, 2008 2:02 PM
To: rblayzor@inoc.net
Cc: cisco-bba@puck.nether.net
Subject: Re: [cisco-bba] MTU challenges on Cisco 7206VXR

Thanks for the feedback. So, do you think that 'mtu 1504' is a valid
configuration item and I should leave that alone? I would prefer not to
drop 8 bytes from my PPPoA customers if removing the 'mtu 1504' entry would
solve the problem.

Frank

-----Original Message-----
From: Robert Blayzor [mailto:rblayzor.bulk@inoc.net]
Sent: Thursday, March 13, 2008 7:39 AM
To: frnkblk@iname.com
Cc: cisco-bba@puck.nether.net
Subject: Re: [cisco-bba] MTU challenges on Cisco 7206VXR

Frank Bulk - iNAME wrote:
> Ok, I agree. So where are the problems coming from?
>
> I do see that on the "root" ATM interface I have the following entered:
> interface ATM4/0
> mtu 1504
> no ip address
> no atm ilmi-keepalive
> !
> And the PVCs are defined under a sub-interface as follows:
> interface ATM4/0.100 multipoint
> pvc 0/261
> encapsulation aal5mux ppp Virtual-Template1
> !
> pvc 0/262
> encapsulation aal5snap
> protocol pppoe
> !
> pvc 0/263
> encapsulation aal5mux ppp Virtual-Template1
> !
>
> Is it possible that the PVC connections are inheriting the 1504 for the
> ATM4/0 interface and that I should delete that line altogether?


I'd leave the MTU at default on the ATM interface and change the MTU on
your virtual-template to 1492. We have tens of thousands of
DSL/Ethernet subs using PPPoE/oA with no MTU problems at all. Usually
the device terminating the PPPoX is smart enough to know to adjust the
MTU... some not so much.

There are some devices in which you have to clamp the mss because the
client can't figure it out for some reason... however, that should be
few and far between.


--
Robert Blayzor
INOC
rblayzor@inoc.net
http://www.inoc.net/~rblayzor/

> SELECT * FROM users WHERE clue > 0
0 rows returned

_______________________________________________
cisco-bba mailing list
cisco-bba@puck.nether.net
https://puck.nether.net/mailman/listinfo/cisco-bba

_______________________________________________
cisco-bba mailing list
cisco-bba@puck.nether.net
https://puck.nether.net/mailman/listinfo/cisco-bba
Re: MTU challenges on Cisco 7206VXR [ In reply to ]
Let me expand the question:

What combination of:
mtu xxxx
ip tcp adjust-mss yyyy
ppp mtu adaptive
should I be using on a Virtual Template? I have none of them today, and
only with PPPoE do things break sometimes.

Frank

-----Original Message-----
From: cisco-bba-bounces@puck.nether.net
[mailto:cisco-bba-bounces@puck.nether.net] On Behalf Of Frank Bulk
Sent: Thursday, November 06, 2008 9:36 AM
To: rblayzor@inoc.net; cisco-bba@puck.nether.net
Subject: Re: [cisco-bba] MTU challenges on Cisco 7206VXR

Should I be applying
ip tcp adjust-mss 1492
to the Virtual-Template, or
mtu 1492

I have a potential maintenance window tonight and I would like to sneak this
in there.

Frank

-----Original Message-----
From: cisco-bba-bounces@puck.nether.net
[mailto:cisco-bba-bounces@puck.nether.net] On Behalf Of Frank Bulk
Sent: Thursday, March 13, 2008 2:02 PM
To: rblayzor@inoc.net
Cc: cisco-bba@puck.nether.net
Subject: Re: [cisco-bba] MTU challenges on Cisco 7206VXR

Thanks for the feedback. So, do you think that 'mtu 1504' is a valid
configuration item and I should leave that alone? I would prefer not to
drop 8 bytes from my PPPoA customers if removing the 'mtu 1504' entry would
solve the problem.

Frank

-----Original Message-----
From: Robert Blayzor [mailto:rblayzor.bulk@inoc.net]
Sent: Thursday, March 13, 2008 7:39 AM
To: frnkblk@iname.com
Cc: cisco-bba@puck.nether.net
Subject: Re: [cisco-bba] MTU challenges on Cisco 7206VXR

Frank Bulk - iNAME wrote:
> Ok, I agree. So where are the problems coming from?
>
> I do see that on the "root" ATM interface I have the following entered:
> interface ATM4/0
> mtu 1504
> no ip address
> no atm ilmi-keepalive
> !
> And the PVCs are defined under a sub-interface as follows:
> interface ATM4/0.100 multipoint
> pvc 0/261
> encapsulation aal5mux ppp Virtual-Template1
> !
> pvc 0/262
> encapsulation aal5snap
> protocol pppoe
> !
> pvc 0/263
> encapsulation aal5mux ppp Virtual-Template1
> !
>
> Is it possible that the PVC connections are inheriting the 1504 for the
> ATM4/0 interface and that I should delete that line altogether?


I'd leave the MTU at default on the ATM interface and change the MTU on
your virtual-template to 1492. We have tens of thousands of
DSL/Ethernet subs using PPPoE/oA with no MTU problems at all. Usually
the device terminating the PPPoX is smart enough to know to adjust the
MTU... some not so much.

There are some devices in which you have to clamp the mss because the
client can't figure it out for some reason... however, that should be
few and far between.


--
Robert Blayzor
INOC
rblayzor@inoc.net
http://www.inoc.net/~rblayzor/

> SELECT * FROM users WHERE clue > 0
0 rows returned

_______________________________________________
cisco-bba mailing list
cisco-bba@puck.nether.net
https://puck.nether.net/mailman/listinfo/cisco-bba

_______________________________________________
cisco-bba mailing list
cisco-bba@puck.nether.net
https://puck.nether.net/mailman/listinfo/cisco-bba

_______________________________________________
cisco-bba mailing list
cisco-bba@puck.nether.net
https://puck.nether.net/mailman/listinfo/cisco-bba
Re: MTU challenges on Cisco 7206VXR [ In reply to ]
> -----Original Message-----
> From: cisco-bba-bounces@puck.nether.net
> [mailto:cisco-bba-bounces@puck.nether.net] On Behalf Of Frank Bulk
> Sent: 06 November 2008 16:38
> To: cisco-bba@puck.nether.net
> Subject: Re: [cisco-bba] MTU challenges on Cisco 7206VXR
>
> Let me expand the question:
>
> What combination of:
> mtu xxxx
> ip tcp adjust-mss yyyy
> ppp mtu adaptive
> should I be using on a Virtual Template? I have none of them
> today, and
> only with PPPoE do things break sometimes.

ip tcp adjust-mss should be about 40 less than the max MTU you can
support (20 for the ip header, and 20 for the tcp).

We have to support a MTU of 1460 for some pppoa clients due to various
l2tp suppliers and access routers that don't support mtu > 1500. We use
a number of virtual-templates depending on the l2tp supplier.

Can be a bit of a headache sometimes - great tool for clients to
diagnose issues is mturoute:

http://www.elifulkerson.com/projects/mturoute.php

Jonty.
_______________________________________________
cisco-bba mailing list
cisco-bba@puck.nether.net
https://puck.nether.net/mailman/listinfo/cisco-bba
Re: MTU challenges on Cisco 7206VXR [ In reply to ]
Thanks for all the feedback on and offlist. I put
mtu 1492
ip tcp adjust-mss 1452
in production during a maintenance window this morning and things have been
working fine.

Regards,

Frank

-----Original Message-----
From: cisco-bba-bounces@puck.nether.net
[mailto:cisco-bba-bounces@puck.nether.net] On Behalf Of Jonty Bale
Sent: Tuesday, November 11, 2008 5:57 AM
To: cisco-bba@puck.nether.net
Subject: Re: [cisco-bba] MTU challenges on Cisco 7206VXR


> -----Original Message-----
> From: cisco-bba-bounces@puck.nether.net
> [mailto:cisco-bba-bounces@puck.nether.net] On Behalf Of Frank Bulk
> Sent: 06 November 2008 16:38
> To: cisco-bba@puck.nether.net
> Subject: Re: [cisco-bba] MTU challenges on Cisco 7206VXR
>
> Let me expand the question:
>
> What combination of:
> mtu xxxx
> ip tcp adjust-mss yyyy
> ppp mtu adaptive
> should I be using on a Virtual Template? I have none of them
> today, and
> only with PPPoE do things break sometimes.

ip tcp adjust-mss should be about 40 less than the max MTU you can
support (20 for the ip header, and 20 for the tcp).

We have to support a MTU of 1460 for some pppoa clients due to various
l2tp suppliers and access routers that don't support mtu > 1500. We use
a number of virtual-templates depending on the l2tp supplier.

Can be a bit of a headache sometimes - great tool for clients to
diagnose issues is mturoute:

http://www.elifulkerson.com/projects/mturoute.php

Jonty.
_______________________________________________
cisco-bba mailing list
cisco-bba@puck.nether.net
https://puck.nether.net/mailman/listinfo/cisco-bba

_______________________________________________
cisco-bba mailing list
cisco-bba@puck.nether.net
https://puck.nether.net/mailman/listinfo/cisco-bba