[rbak-nsp] MLPPP with 8 X DSL on SE100; packet loss upstream

[Lucy West]

Hello - wondering if anyone's seen/done something similar....
We've got a Cisco CPE - 8 x SDSL MLPPP'd into an SE100 which is acting as an LNS.
SE100 is running SEOS 6.5.1.3.
The traffic is sent at 7111pps upstream from the CPE; on the SE100 we see that the packet counters on the L2TP source interface (i.e.ingress interface) is 7112 pps inbound and 6522 pps on the egress interface (the MPLS interface into the core). Looking at the Ixia's input and output there's about a 50,000 Byte difference between what it sends out (via the CPE LAN) and what it receives.

Has anyone else seen similar packet loss or issues with MLPPP? Anyone got it working correctly?!
We've also been able to crash the box sending ~15Mbps traffic with one CPE attached with 8X SDSL lines MLPPP'ed.

Cheers
Lucy



-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://puck.nether.net/pipermail/redback-nsp/attachments/20110914/62d21c36/attachment.html>