What steps will reproduce the problem?
I am unsure of the exact conditions which cause this issue, but I can
reasonably reliably reproduce it with four nodes daisy-chained (nodes
restricted so they can only hear their direct neighbors).
Take the scenario A-B-C-D, then remove B temporarily, before bringing it back
in. Inspect the routing table of node C, and you might see something like the
following:
$ nc6 -u fd00:b81e:1::4 2000
route
key destination gateway iface
2 fd00:b81e:1::5/128 fe80::f8de:ff:fe00:5 pan
3 fd00:b81e:1::3/128 fe80::f8de:ff:fe00:3 pan
4 fd00:b81e:1::4/128 fe80::f8de:ff:fe00:5 pan
1 ::/0 fe80::f8de:ff:fe00:3 pan
^C
Here node A has address x::1, B is x::3, C is x::4 and D is x::5. Note that the
routing table on node C contains a route for itself via node D!
When this happens on both nodes C and D, it appears this somehow causes packets
destined for either to be forwarded back and forth, a lot. I don't see why it
should do that even with these routes in place, but it seems to nevertheless.
What is the expected output? What do you see instead?
I would never expect to see a route for a local IP address listed via a remote
node.
What version of the product are you using? On what operating system?
TinyOS 2.x trunk as of a few weeks ago.
Please provide any additional information below.
When enabling the MRHOF feature, the A-B-C-D scenario fails to converge
properly, routing wise. I haven't investigated that further as we can live
without that feature, but I was seeing a lot route-for-own-IP-via-remote-node
entries with MRHOF enabled.
Original issue reported on code.google.com by jmatts...@dius.com.au on 20 Dec 2011 at 11:44
Original issue reported on code.google.com by
jmatts...@dius.com.au
on 20 Dec 2011 at 11:44