3 ms·
AT&T regularly assigns my phone an IP in 10/8, instead of using 100.64/16 as they should [1]. IIRC, they used to even have the gall to use 172.16/12, which is c
by nickysielicki 9y ago
AT&T regularly assigns my phone an IP in 10/8, instead of using 100.64/16 as they should [1]. IIRC, they used to even have the gall to use 172.16/12, which is crazy when you consider the amount of corporate networks using those addresses.
This caused issues where my phone would try to get on wifi, but the DHCPACK would be sent along on the existing interface rather than the one coming up. So the wifi icon was continually bouncing back and forth. My only solution was to go into airplane mode and bring the cellular down before bringing the wifi up. I don't think Android ever addressed this issue, and I had to switch around the entire subnet to avoid the conflicts.
If I knew enough about how Android worked, I'd write a patch to have all android interfaces in their own linux netns, with the dhcp client exec'd in that netns, that way you'd never have to worry about this sort of conflict.
[1]: https://tools.ietf.org/html/rfc6598 https://tools.ietf.org/html/rfc6598
- isaack 9y agoA carrier deploying NAT might not be using CGNAT, and hence, in compliance with the spec, decide not to use the 100.64/16 space. One requirement of CGNAT, for instance, requires that the carrier's router be able to handle "address crashes". [1] [1] https://tools.ietf.org/html/rfc6598#section-4 https://tools.ietf.org/html/rfc6598#section-4
- nickysielicki 9y agoI think it says that you can use it even if you're not doing CGNAT, provided that you are a service provider. > In particular, Shared Address Space can only be used in Service Provider networks OR on routing equipment that is able to do address translation across router interfaces when the addresses are identical on two different interfaces. (edit: no, very clearly, "Devices MUST be capable of performing address translation when identical Shared Address Space ranges are used on two different interfaces." ) Also, is it wrong to assume that cellular networks are able to handle "address crashes" due to the inherent centralization that comes as a result of having clients maintain the same IP (and same connections) as the device hops from tower to tower? Maybe I don't understand the topics at play here...