[OLSR-users] Re: OLSR Network Dying

[John Gorkos]

On Tuesday 28 June 2005 10:52 pm, Andrew Hodel wrote:
> Any ideas what is wrong, entire network is down...
>
> Andrew
>
Are you using any plugins?  Does every node have an identical olsrd.conf?  I 
had some similar problems until I took out all of the plugins, got everyone 
on the same version of olsrd, and used identical olsrd.conf files across the 
entire network.  I also had two "key" nodes (one was an HNA, one was not) 
that would do exactly what you're describing, so I set up a cron for them to 
reboot every night at 0200.  That was the final solution, and my network 
reliability has skyrocketed (save for the fact that I would _NEVER_ use 
WRT54Gs in a commercial environment again).

I'm attaching my "common" olsrd.conf from one of my nodes.  I use hysterisis, 
not LQ.  Nothing against LQ or Thomas' work, it's just what I started with 
and I was having enough success to not mess with it.  The olsrd.conf below is 
in place on every node in my network, save for the two HNAs, which use the 
dyn-gw plugin.  My network addressing scheme is the 192.168.7.0/24 subnet:
i.e. inet addr:192.168.7.43  Bcast:192.168.7.255  Mask:255.255.255.0

#
# olsr.org OLSR daemon config file
#
DebugLevel      0
IpVersion       4

Hna4
{
}

Hna6
{
}

AllowNoInt      yes

IpcConnect
{
     MaxConnections  1
     Host            127.0.0.1
}

UseHysteresis   yes
HystScaling     0.10
HystThrHigh     0.60
HystThrLow      0.20
Pollrate        0.15

Interface "eth1"
{
  HelloValidityTime       20.0
  TcValidityTime 40.0
}


> On 6/28/05, Andrew Hodel <(spam-protected)> wrote:
> > In our network with 27 nodes, the network will simply die.
> > Interesting is that olsr continues to run on each node, and each node
> > is recieving packets from all other nodes.  Every node shows every
> > other node in the topology and all routes are in place.  Whenever I
> > try to do anything over the network however, ping tcp whatever,
> > nothing works.
> >