3905 losing registration VDSL2 NBN

[nathanhutchinson]

3905 (CP3905.9-4-1SR3) to CUCM (14.0.1.13900-155) via DMVPN, phones registration constantly restarting when backhaul to site is via VDSL2 (NBN FTTN) but stable when via Cellular (Telstra). 3905 is connected via a Cisco IE-4010-16S12P but we are also having the same issue with other sites running different Cisco hardware.

To first common point in core network

Via VDSL: MTU 1400 round-trip min/avg/max = 52/54/56 ms

Via Cellular: MTU 1464 round-trip min/avg/max = 49/52/60 ms

Have tried to set 3905 MTU to 1400 via CUCM and DHCP option 26 though not sure if either is working as debug ip_addr command run on 3905 shows:

2: eth0: <BROADCAST,MULTICAST,UP,10000> mtu 1500 qdisc pfifo_fast qlen 1000

fox-cli logs from 3905 attached

[b.winter]

Moved your post to the networking forum, as this has nothing to do with collaboration.

[balaji.bandi]

is this ever worked ? or not working from the day installed ?

Other than Phone everything work ok ? is the Trunnel stable ?

can you post the configuration and IOS code running ?

what Phone models ? 

what Logs you see on the Hub side ?

[nathanhutchinson · Accepted solution]

after a lot of troubleshooting and packet captures it was found that we needed to put "ip tcp adjust-mss 1300" on the interface within the vpnv4 as packets from these networks were too large for the backbone. Note applying the tcp adjust-mss to the backbone interface doesn't resolve the issue.