PPP multilink across 2 T1's, unbalanced loading all of the time

[beggert]

I'm looking for isolation suggestions or solutions. I have a 7513 using a VIP2 connected to a 2600 router. It appears that MPPP is not load balancing correctly. The ratio is between 10:1 and 20:1 on the output of the 7513 toward the 2600. The 7513 is running 12.2.13. The configuration is used for other routers using MPPP and those routers do not have this unbalanced issue. The serial interfaces were reset and process switching was attempted, but the ratio remained the same. I've read that other's had strange problems and they completely removed MPPP, rebooted and reconfigured MPPP. Another option is to move MPPP off of the VIP2 to the 8 port serial card to see if the VIP2 has a problem. Below is the MPPP configuration for the 7513. Being load balancing is done on the output, I believe the issue is with this box.

ip cef distributed

interface Multilink1

bandwidth 3056

ip address 10.10.10.1/30

no ip route-cache

ip ospf message-digest-key 70 md5

ip ospf database-filter all out

no ip mroute-cache

no cdp enable

ppp multilink

no ppp multilink fragmentation

multilink-group 1

interface Serial0/0/0

no ip address

encapsulation ppp

no ip route-cache

no ip mroute-cache

no fair-queue

serial restart-delay 0

ppp multilink

multilink-group 1

!

interface Serial0/0/1

no ip address

encapsulation ppp

no ip route-cache

no ip mroute-cache

no fair-queue

serial restart-delay 0

ppp multilink

multilink-group 1

[tepatel]

I don't think that VIP2 has any problem.

First need to know the reason for "no ppp multilink fragmentation" under the "interface multilink 1"..You need fragmentation on.

With fragmentation on, need to look on "sh interface accounting" on the traffic origination or sending router so that we can see how many packets are sent on the member links.

[rbristow]

You may want to try CEF rather than DCEF.

Regards,

Rob Bristow

CCIE #3335

[beggert]

The network has other routers using the same configuration for ppp multilink and they do not have this problem. Using the same configuration, the working members have fairly equal packets sent. On this box, one t1 has 22K out and the other has 5M out. We're going to move the 2 T1's to available serial ports to test, afterwards we'll return the T1's to the VIP2.

[beggert]

found work around: other messages said to remove multilink, reload, reconfig multilink to restore MPPP. I was not able to reload the router, so the multilink was removed and a new multilink using a different identifier was configured. By using a different identifier, we no longer had any reference to the previous multilink. By doing this, the load balancing issue was resolved without reloading the router. Also, bug ID CSCEA56957 could be related to this issue and a recommendation to upgrade to 12.2.16 from 12.2.13 was given.