Currently supernodes are great at forwarding packets that are not meant for them within the above range towards batman interfaces based on the destination.
eg.
2a02:790:ff:700::/64 dev bat0 proto kernel metric 256 pref medium
responses of these are reaching the sn via batX and are marked as such iirc.
Packets that are within the above range but originated on the sn itself are not routed properly, but routed towards eth0 (default).
This leads to marsian packets, as requests may enter a supernode via e.g. gre-leintor and leave it through eth0.
I called @lemoer earlier today; and evaluated afterwords:
the source address for response packets is something like 2a02:790:ff:700::7001 while the address of usual requests towards the internet is 2001:41do:...
I think a source based routing for stuff coming out of the title-prefix would solve this issue.
Anything that sources 2a02:790:ff::... should be routed towards gre-leintor, right?
AiyionPrime
commented
3 years ago
Currently supernodes are great at forwarding packets that are not meant for them within the above range towards batman interfaces based on the destination.
eg.
2a02:790:ff:700::/64 dev bat0 proto kernel metric 256 pref mediumresponses of these are reaching the sn via batX and are marked as such iirc.
Packets that are within the above range but originated on the sn itself are not routed properly, but routed towards eth0 (default). This leads to marsian packets, as requests may enter a supernode via e.g. gre-leintor and leave it through eth0.
I called @lemoer earlier today; and evaluated afterwords: the source address for response packets is something like
2a02:790:ff:700::7001while the address of usual requests towards the internet is2001:41do:...I think a source based routing for stuff coming out of the title-prefix would solve this issue. Anything that sources
2a02:790:ff::...should be routed towards gre-leintor, right?