[lime-users] Internet traffic not going through

Nk nk at os.vu
Mon May 15 02:10:08 UTC 2017


Hi Gui

Thank you for your reply. Please see my update from a few hours back for more info.

1] Yes that means boot sorry

2] Do you mean those three steps as actually flashing the routers from stock lede [not-lime] and leaving them on directly or do you mean power on for the first time after a successful flash, exposure to the network, and then power off, and back on?

3] I mean [B]’s luci, the one to which the client is attached.

I’ve actually found that /etc/init.d/network restart does fix the issue. I’ve added sleep 60 && /etc/init.d/network restart in rc.local so that every router restarts its network service after 60 seconds from boot and everything seems to work stably whatever I do, but I’ll need to do some further testing to be sure.

[P.S.: since you’ve been so nice to reply, to avoid opening a new thread and being targeted by all of the ml’s hate ahah ;] can I ask you what ratemax means under Network > BMX6 > Configuration > Interfaces? I think I’ve figured out that those are kbps and not bps, but I don't believe there would be a hard-coded upper cap to how fast wireless meshing would be allowed to work, right? Raising all of the nodes to 450000 [as in theoretical max for 3x3 MIMO] seems to get me faster speeds, but on a sample size of 10 tests, so not statistically significant, and probably biased by my obsession with speed and related hopes of maxing them out ;]

Thank you so much again!

Nk

On May 15, 2017, 4:00 AM +0200, Gui Iribarren <gui at altermundi.net>, wrote:
> On 13/05/17 13:22, Nk via lime-users wrote:
> > Hi
> >
> > I have a strange issue where sometimes, seemingly after a new gateway
> > appears on the network, other routers that were offline during the
> > change, cannot route internet traffic correctly even though BMX tunnels
> > shows everything as being perfect, including the inet4 line with the
> > exit node saying *internet*.
> >
> > Steps to reproduce:
> >
> > 1] set up a node with internet access [A]
> > 2] connect client via ethernet to a known to work second [and non-exit]
> > node [B] which is wirelessly meshing with [A]
> > 3] ping internet from client [it works]
> > 4] shut down [A]
> > 5] shut down [B]
> > 6] turn on a new node [C] with the exact same config as [A] and connect
> > it to the same WAN
> > 7] turn [B] back on
> >
> > Everything in luci will appear perfectly, you’ll be able to ipv6 ping
> > any node, but any internet bound traffic will time out, even from the
> > luci diagnostics page [not a client routes issue].
> >
> > After rebooting [B], everything works consistently across reboots.
>
> i think i don't understand. at point 7], "turn back on" means boot,
> right? so what you mean is that it doesn't work during the "first" boot,
> but it works ok from the "second" boot and further??
> what happens if you do, "from scratch":
> 1) power on C
> 2) power on B
> 3) power on A
>
> does everything work?
>
> and when you say that it doesn't work "from luci", which luci are you
> referring to? C, B or A luci?
>
> cheers!
>
> gui
>
> >
> > I’ve yet to try with standard 1607, but I’m using an extremely lightly
> > modified config. I’ll do that tomorrow anyway.
> >
> > Is there a way to restart batman to see if that’s the issue? [can’t find
> > its entry in “startup”] Is there a way to restart some other service to
> > figure out what the issue is? What should I look for in logs?
> >
> > Is there a general way to force-reset something if a node sees a mesh
> > exit-node but somehow isn’t able to route traffic to it?
> >
> > Thank you
> >
> > Nk
> >
> >
> >
> > _______________________________________________
> > lime-users mailing list
> > lime-users at lists.libremesh.org
> > https://lists.libremesh.org/mailman/listinfo/lime-users
> >
> _______________________________________________
> lime-users mailing list
> lime-users at lists.libremesh.org
> https://lists.libremesh.org/mailman/listinfo/lime-users
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.libremesh.org/pipermail/lime-users/attachments/20170515/0fa592ec/attachment.html>


More information about the lime-users mailing list