← Back to team overview

maas-devel team mailing list archive

Re: Cobbler Replacement (?)

 

Hi, Julian,

Thanks much for your prompt response!  I've seen it evolved from Orchestra to MAAS, and it's getting much simpler.  And now, I love more the direction of making it less dependent to other tools.  Very promising product. Thanks guys!  

I'd like to provide some feedback based on my preliminary experiences:

The current state is somehow hit-n-miss.  For example, right after 12.04 was released, I got it working within VirtualBox - vm01 for maas and juju, vm02 for zookeeper agent node, vm03 for mysql node, vm04/05 for wordpress, vm06 for haproxy.  But now, after I teared down all the juju service nodes (i.e., destroyed all the services, terminated all the machines), and tried to redo it again, I got an unstable MAAS environment as described below:

* The juju service node (say, mysql) deployed, but the agent-state is kept in "not-started" status, 
* Rebooted the node, say vm03 via PXE, it showed up in the MAAS dashboard and the status turned from "Ready" to "Allocated to root" 
* The cloud-init has some output along with the ssh public key info on the console
* Now, I'm stuck.  When I tried to ssh into the node and take a look at the juju logs on the service node, I  got a prompt asking for password, and I don't know what the password is. 

Any ideas what I need to do to reestablish the link between juju and the node agent?

The second feedback is around the enlisting via the PXE booting without entering the MAC address to the MAAS dashboard first.  

Scenario #1: Enter the MAC address through MAAS dashboard first, then PXE-boot the node:

It went through the booting process really quick and the node is shut down.  There's no OS install on the node.  The node status on the Dashboard turned into Commissioning status.

Scenario #2: Without entering the MAC address in the MAAS dashboard, PXE-boot the node:

The node gets booted, and automatically enters into the enlisting install.  In this case, OS was actually installed on the node.  Then the node turned into Declared status.  

Question: Can the system skip the OS install for the enlisting process in Scenario #2?  That would make it consistent with Scenario #1.

The third feedback: During the OS install (booted through PXE via MAAS), I had to manually answer " Continue without a default route?" in the "Configure the network" screen.  After selecting "Yes" and hitting the Enter key, I then had to manually enter the answer for the "Name server addresses:" prompt in the "Configure the network" screen.  Can you let me know which preseed file I need to edit to automate these two screen?

Best Regards,
Dafang

Follow ups

References