Open Stack

Hi all,

I observed an error in documentation regarding the configuration needed to be done on additional compute-nodes in order to allow vnc access to instances.
>From what I understood from doc you just have to copy nova.conf on additional nodes.
However there are some changes to made in the nova.conf of the node you want to add or the cloud.
The parameters to be changed are:--vncserver_proxyclient_address;--vncserver_listen

If on the cloud master they have the values:

--vncserver_proxyclient_address=$ip_cloud_master
--vncserver_listen=$ip_cloud_master

On the newly added compute node they have to be:
--vncserver_proxyclient_address=$ip_compute_node
--vncserver_listen=$ip_compute_node
>From my testing and small understanding ....:) these values are as such:
-vncserver_listen is the address that you will find in libvirt.xml corresponding to the instance started on this server. If you put a value different from the ip addresses on this server the instances won't come up.
-vncserver_proxyclient_address is the address that nova-consoleauth will associate with the requested token.

I don't know if I explained clearly enough so I will give an example.
I have 2 servers in my cloud, node01(master_cloud), node02(compute_node). If on node02 the value vncserver_proxyclient_address is $ip_cloud_master after I start an instance on node02 vnc will direct me to $ip_cloud_master:5900 which is an old instance created on node01. If If on node02 the value vncserver_proxyclient_address is $ip_compute_node
 after I start an instance on node02 vnc will direct me to 
$ip_compute_node:5900 which is what I am looking for. BINGO!!

Regards,
Gabriel
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack/attachments/20120518/ca1204d5/attachment.html>