Closed bephinix closed 5 years ago
@robbertkl It even simplifies the code base!
@robbertkl I had to change two rules for internal networks.
If use the same dropping rules for internal IPv6 networks as we do for IPv4, we will end with these rules:
0 0 DROP all * dckrNet1 !fddd:0:0:11::/64 ::/0
0 0 DROP all dckrNet1 * ::/0 !fddd:0:0:11::/64
This also blocks Neighbor Discovery Protocol (counterpart to ARP for IPv4) so ICC will not be possible - even when it is enabled. As a solution we should not check for a specific address but for input and output interfaces:
0 0 DROP all dckrNet1 !dckrNet1 ::/0 ::/0
0 0 DROP all !dckrNet1 dckrNet1 ::/0 ::/0
With these rules NDP packets will not be dropped and if you can still decide with the ICC flag if you want to allow ICC.
Commit: 4fba810337a632db042e8622e30760b8299cd1cc
This MR implements an IPv6 version of docker/libnetwork#2117.
Fixes #22 Closes #22
/cc @robbertkl