[quagga-users 14014] Re: Running Multiple Instances of Quagga

Jafar Al-Gharaibeh jafar at atcorp.com
Wed Jun 17 15:35:12 BST 2015

Thank you Olivier.


On 6/16/2015 9:28 AM, Olivier Tilmans wrote:
> Hello Jafar,
> If you don't have a 'main' process in the network namespace in charge of
> spawning all other processes for that namespace to which you can send
> command, then you will need to remount /var/run each time you issue an
> ip netns exec as mount namespaces are process-inherited.
> e.g.
> mkdir -p /tmp/newns/var/run
> ip netns add newns
> ip netns exec newns bash -c 'mount --bind /tmp/newns/var/run /var/run &&
> command_to_start_quagga'
> ## Later on ...
> ip netns exec newns bash -c 'mount --bind /tmp/newns/var/run /var/run &&
> vtysh'
> A less constraining solution would be to patch the daemons & vtysh to
> add a launch argument to specify to location of the vtysh socket ...
> Best,
> Olivier.
> On 15/06/15 22:10, Jafar Al-Gharaibeh wrote:
>> HI Olivier,
>>     In my case, I have a number of programs that continuously issuing
>> vtysh commands based on state/variables. I have two instances of Quagga
>> running, one in the the global network namespace, and the other in a new
>> network namespace call it newns. In the places where I want to
>> communicate with Quagga in newns, I added "ip netns exe newns" before
>> the vtysh command, but that didn't do the trick. What I hear  you saying
>> is that if I do the mount trick, my command would work (connect to the
>> correct Quagga). is that correct?
>> Thanks,
>> Jafar

More information about the Quagga-users mailing list