[Netkit.users] Troubles wih Quagga/RIP

[Massimo Rimondini]

Dear Vsevolod,

we are aware of this bug. The reason why it is still there is that we
are working on a new release that also involves the core and the kernel,
and this consumes most of the (limited) development time.

A fix to this will be available in the next version of the filesystem
image. In the meantime, you can work this around by proceeding as follows.
- Start a virtual machine that permanently writes on the filesystem image:
vstart vm -W -m 64
- Inside the virtual machine, edit the file /etc/init.d/quagga (or
zebra, they are just the same file) and change line 24 ('echo
"/var/run/quagga/$1.pid"') into 'echo "/var/$1.pid"'.
- Gracefully shut down the virtual machine by typing 'halt' inside it.

This should solve your problem.

Sorry for the inconvenience and please feel free to come back if you
experience further issues.

Regards,
Massimo.


Vsevolod Krishchenko wrote:
> Hi everybody,
>
> I tried Netkit (and UML) for the first time today. They are great, but 
> I got some problems with quagga.
>
> 1. How to stop it? '/etc/init.d/zebra stop' does not work:
>
> r1:~# ps -A | egrep '(ripd|zebra)'
>   479 ?        00:00:00 zebra
>   483 ?        00:00:00 ripd
> r1:~# /etc/init.d/zebra stop
> Stopping Quagga daemons (prio:0): (ripd) (zebra) (bgpd) (ripngd) 
> (ospfd) (ospf6d) (isisd) (ldpd).
> Removing all routes made by zebra.
> r1:~# ps -A | egrep '(ripd|zebra)'
>   479 ?        00:00:00 zebra
>   483 ?        00:00:00 ripd
>
> /etc/init.d/quagga does the same (zebra is just a symlink)
>  
> 2. After ifdown and ifup quagga still thinks that interface is down:
>
> r1:~# ifdown eth2
> r1:~# tail -1 /var/log/quagga/ripd.log
> 2008/10/17 22:27:34 RIP: interface delete eth2 index 5 
> r1:~# ifup eth2 && sleep 30
> r1:~# tail -2 /var/log/quagga/ripd.log
> 2008/10/17 22:27:34 RIP: interface delete eth2 index 5 
> 2008/10/17 22:27:56 RIP: rip_read: cannot find interface for packet 
> from 192.168.20.2 port 520 
>
> So, quagga should be restarted after interface is up. But zebra 
> restarting is a bit tricky because '/etc/init.d/zebra restart' does 
> not work (of course we can just kill, but I don't wanna do it)
>
> 3. Is there a betterway to model network problem? Disabling interface 
> via vconfig/uml_mconsole does not work because interface is up :( 
> ifdown works fine, but ifup does not affect quagga at all  :(
>
> 4. Maybe something is wrong with my host? It's amd64 debian sid, but 
> my little problems does not look like host-related. I can try i386 
> system tomorrow, anyway.
>
>