Hello Pramod,
hardware based fragmentation is supported only if the MLPPP has only one member according to SIP and SPA configuration guide
http://www.cisco.com/en/US/docs/interfaces_modules/shared_port_adapters/configuration/6500series/76cfgct3.html#wp1168724
Hardware MLPPP LFI Guidelines
LFI using MLPPP will function only in hardware if there is just one member link in the MLPPP bundle. The link can be a fractional T1 or full T1. Note the following guidelines:
•
The ppp multilink interleave command must be configured to enable interleaving.
•
Only three fragmentation sizes are supported: 128 bytes, 256 bytes, and 512 bytes.
•
Fragmentation is enabled by default, and the default size is 512 bytes.
•
A policy map having a priority class must be applied to the main interface.
•
When hardware-based LFI is enabled, fragmentation counters are not displayed.
in your case we see that there two members that are two T1 derived from the same channelized T3 port.
Member links: 2 active, 0 inactive (max not set, min not set)
Se3/0/0/5:0, since 12w6d, no frags rcvd
Se3/0/0/7:0, since 3w0d, no frags rcvd
the show provides information that is in contrast:
7733426 lost fragments
but no frags received on member link, meaning other side is not fragmenting, so where are lost these fragments in sending?
in your case using two member links LFI is software based on the SPA and this can be the cause of lost fragments but I have some doubts if the counters are real or not. Are you experiencing a real issue?
An FTP file transfer performs poorly over the bundle?
>> Distributed fragmentation on. Fragment size 512. Multilink in Hardware.
Hope to help
Giuseppe