Does Verizon network apply an idle/inactive timeout threshold for the ipv6 configuration?

Hi community,

In my testing I found that Verizon seems to adopt an idle timeout threshold on its ipv6 network.

In the first place, the ModemManager’s simple connect can setup a bearer successfully, which provides both ipv4 and ipv6 configurations.

Tcpdump revealed that there are ipv6 NDP traffic between the cellmodem’s slaac address, link-local address and the gateway’s link-local address, about 15m or so. Also the cellmodem interface receives ipv6 RA.

Despite of these existing ipv6 traffic, if I leave the cellmodem interface idle, for about 3-4 hours, the ipv6 configuration will be dropped. Although I could simple connect again and this time there will be new/different slaac address, link local address on cellmodem and new gateway’s link-local address, the ipv6 configuration doesn’t really work - I can’t even ping gateway’s link-local address from the cellmodem interface

Further testing found that if I send a ping6 packet from the cellmodem to the gateway’s link-local address periodically, say once per hour, would help keep alive the initial ipv6 configuration to prevent it from being dropped anymore.

Does anyone run into anything similar? Does Verizon enforces ipv6 idle timeout threshold at all?

Thanks!
Harry

which module are you talking about?
Is problem only happening in Verizon network?
How about T-MOBILE?

Hello,

I am using MC7354 with vzwinternent APN, so far this symptom was observed with Verizon and I have no idea about T-mobile.

Thanks,
Harry

is that no problem is found in IPV4?

I think this could be a good workaround, right?

The vzwinternet APN provides both ipv4 and ipv6 configurations. The ipv4 configuration doesn’t need such keepalive traffic and every hour the cellmodem interface receives the dhcp renew event with the same ipv4 settings.

Yeah, the overhead of extra keepalive ping6 traffic is really trivial, given existing ipv6 NDP and RA traffic. I will keep what I have found, just would like to consult the community if anyone else may also run into sth similar and/or have better solutions.

Thanks again,
Harry