I'm still not quite sure why your MAAS provisioning doesn't have the prompt 
without a default route if your bare metal only has one NIC.  Your bare metal 
install must somehow have found a default route in the network settings, I 
guess.

I used a dedicated node mainly is because I didn't know it would be possible to 
use the MAAS/juju node as the zookeeper node.  From this MAAS wiki page 
- https://wiki.ubuntu.com/ServerTeam/MAAS/Juju, it requests to have at least 
two nodes added to MAAS to start to use juju.  To my understanding, one node is 
for Zookeeper node, another is for any possible juju service node.  The 
Zookeeper node was automatically provisioned by MAAS (well, except the manual 
answer to the network configuration screen due to the absence of a default 
route, and also, the manual answer to the name server, which is the IP address 
of the MAAS/juju node).  Can you share with me how you managed to have 
maas/juju/zookeeper loaded in one node?

Best,
Dafang


________________________________
 From: Sébastien Han <[email protected]>
To: Dafang Zhang <[email protected]> 
Cc: Julian Edwards <[email protected]>; 
"[email protected]" <[email protected]> 
Sent: Sunday, May 6, 2012 3:36 PM
Subject: Re: [Maas-devel] Cobbler Replacement (?)
 

Ok, I see.
I also have one issue with the preseed file in my bare metal setup.

May  2 14:41:00 ubuntu partman:   No matching physical volumes found
>May  2 14:41:00 ubuntu partman:   No volume groups found
>May  2 14:41:00 ubuntu partman:   Reading all physical volumes.  This may take 
>a while...
>May  2 14:41:00 ubuntu partman-lvm:   No volume groups found

During the installation I have to manually choose a disk to install the 
system...

My bare metal setup is mainly the same as the virtual setup except that I only 
use one NIC for the MAAS node.
But the weirdest think is that I have the exact same trace in the node of my 
virtual setup. And I'm not ask for choose a disk in the virtual setup...

One more thing why do you use a dedicated node for zookeeper?

Cheers!



On Sun, May 6, 2012 at 6:10 AM, Dafang Zhang <[email protected]> wrote:

Hi, Sebastien,
>
>Thanks for looking into this.  You are correct that I only enabled the 
>Host-Only interface in VirtualBox, but I did user maas-dhcp package to enable 
>the MAAS node to manage DNS and DHCP.  In a real-world scenario, all the 
>service nodes would sit behind the maas/juju node, so that they will only have 
>one NIC (or a team) to communicate with the MAAS node only, and use the MAAS 
>node's squid as the proxy to communicate with the rest of the world.  I could 
>have the service node VM's to have two NIC's, but I intended to set it up this 
>way to mimic a scenario that would secure the service nodes (in your term, 
>MAASlave nodes). 
>
>When I setup the MAAS node, I put two NIC's on it, one NAT (communicate with 
>the Net), another Host-Only (communicate with the zookeeper node and all other 
>service nodes).  I used a preseed URL to automate the MAAS node base ubuntu 
>install.  I don't have this issue with the MAAS node install.  
>
>So, I only have this issue when I use MAAS to provision the service nodes, 
>each has just one NIC.  I don't know why "setting netcfg/no_default_route" is 
>not working.
>
>Back to your bare metal setup, sounds like you have only 1 private NIC for 
>your bare metal, so you don't have the issue that needs to manually answer the 
>prompts during the OS provisioning?
>
>Best,
>Dafang
>
>
-- 
Mailing list: https://launchpad.net/~maas-devel
Post to     : [email protected]
Unsubscribe : https://launchpad.net/~maas-devel
More help   : https://help.launchpad.net/ListHelp

Reply via email to