Routing Loop Detected

[hgn]

The following pattern is triggered by executing make promotion. This should never occur, and it was the last time triggered in the beginning of DMPR, mmmh

Destination Router is not connected
forward [highest-bandwidth]    8 ->   24
forward [highest-bandwidth]   24 ->   20
forward [highest-bandwidth]   20 ->   25
forward [highest-bandwidth]   25 ->   24
forward [highest-bandwidth]   24 ->   20
forward [highest-bandwidth]   20 ->   25
forward [highest-bandwidth]   25 ->   24
forward [highest-bandwidth]   24 ->   20
forward [highest-bandwidth]   20 ->   25
forward [highest-bandwidth]   25 ->   24
forward [highest-bandwidth]   24 ->   20
forward [highest-bandwidth]   20 ->   25
forward [highest-bandwidth]   25 ->   24
forward [highest-bandwidth]   24 ->   20
forward [highest-bandwidth]   20 ->   25
forward [highest-bandwidth]   25 ->   24
forward [highest-bandwidth]   24 ->   20
forward [highest-bandwidth]   20 ->   25
forward [highest-bandwidth]   25 ->   24
forward [highest-bandwidth]   24 ->   20
forward [highest-bandwidth]   20 ->   25
forward [highest-bandwidth]   25 ->   24
forward [highest-bandwidth]   24 ->   20
forward [highest-bandwidth]   20 ->   25
forward [highest-bandwidth]   25 ->   24
forward [highest-bandwidth]   24 ->   20
forward [highest-bandwidth]   20 ->   25
forward [highest-bandwidth]   25 ->   24
forward [highest-bandwidth]   24 ->   20
forward [highest-bandwidth]   20 ->   25
forward [highest-bandwidth]   25 ->   24
forward [highest-bandwidth]   24 ->   20
drop packet, ttl 0

[reisub-de]

I can't reproduce this right now, which python version/distro?

[hgn]

Yes, I saw it once on the PC next by me. Is the seed somehow configurable? I mean on this local box here it is reproducable, because the seed seems identical for all runs. Probably:

seed=random()
for infty:
   run_simu()
   if loop: break;

I mean by hand ...

[reisub-de]

Yeah, it always uses the same seed to be reproducible, the seed can be given as parameter random_seed_prep and random_seed_runtime in dmprsim/analyze/promo.py#L36

[hgn]

probably the way to trigger this bug, iterating over ...

[reisub-de]

Well, the random behavior depends on the python version ^^, so I would need that ;)

Also, this is probably no bug, just a temporary routing loop which is inevitable when there are moving nodes, it should clear up after a while

[hgn]

It clears up, but I think this MUST be avoided. In real life it will blow up the link until TTL reaches zero. So let it be there for now but I am really skeptical that it can be there in the final version somehow ...

[reisub-de]

It can't be avoided on the dmpr level, maybe somehow in iptables/nftables

[hgn]

Mhh, let's move this topic to some later releases[TM] for now ... ;.)