freifunk-gluon / gluon

a modular framework for creating OpenWrt-based firmwares for wireless mesh nodes
https://gluon.readthedocs.io
Other
553 stars 325 forks source link

Node does not reply to Neighbour Discovery for v6 Nextnode #2854

Closed nrbffs closed 10 months ago

nrbffs commented 1 year ago

Bug report

What is the problem?

Sometimes, the GL-inet XE300 (maybe also other models) does not reply to neighbour discovery requests for its v6 nextnode address. Therefore, ping, status page and ssh do not work over the nextnode address.

From a client, a ping to the node's v6 address on br-client works fine. However, pinging the nextnode does not work. One can see from a packet capture that the node doesn't reply to neighbour discovery requests for the nextnode address.

tcpdump on -i any in promiscious mode on the node shows that the neighbour discoveries actually reach the node. Listening on local-node also shows the neighbour discoveries.

Without promiscious mode (i.e. with tcpdump -npi local-node), the neighbour discovery is not visible on the node.

What is the expected behaviour?

The node responds to neighbour discoveries and replies to pings.

Gluon Version: 157a74c5ab1355ce2d62b53683262fe8e0d11c70

Site Configuration: https://gitlab.freifunk-stuttgart.de/firmware/site-ffs/-/tree/6429e1491dde9a9482d0b067f4b1d2993beacfc4

Custom patches:

maurerle commented 10 months ago

I have not seen this issue for any device yet. In your scenario, the link-local address exists on local-node@local-port interface but the node does not reply to it? Or is the local-node address not set at all?

If this only affects GL-inet XE300 you might want to prefix the title with the model? Or close this issue if this has not been seen recently anymore

nrbffs commented 10 months ago

I have not seen this issue for any device yet. In your scenario, the link-local address exists on local-node@local-port interface but the node does not reply to it? Or is the local-node address not set at all?

The address was configured on local-node@local-port and was reachable from the node itself. Clients were unable to reach it, though.

If this only affects GL-inet XE300 you might want to prefix the title with the model? Or close this issue if this has not been seen recently anymore

I remember having observed this on other models, too. Please give me a bit of time to try and reproduce this. Any ideas which debug information I can provide to help further track this down?

nrbffs commented 10 months ago

Unable to reproduce on XE300 and OR750i with our firmware build 3.0+2024-01-09-g.62dbf788-s.3621e0e (v2023.2). Closing for now.