[chef] Re: Re: Re: Re: Re: Ideas on kickstarting a node and auto-applying a role

[Paul Choi < >]

Hi Gilles,

Actually, I'm really interested in what others have to say about this 
idea, so thanks for your input.

Bryan McLellan responded saying you can do "chef-client -j <json file>" 
to apply a role from a node.
So, in my kickstart script's %post section, I have:
  curl http://<blah>/cgi-bin/chef_role.cgi?<node fqdn> > /tmp/role.json
  chef-client -j /tmp/role.json

And chef_role.cgi is just a little perl script I wrote that takes the 
hostname and spits out role/cluster assignment.

I did think about maybe setting up knife on each host. Or using 
Chef::REST with an admin user's client PEM cert on the node, but I 
thought I'll just do "chef-client -j", since it at least works for me. :)

-Paul


On 5/20/10 10:34 PM, Gilles Devaux wrote:
> Ouch, old thread, sorry.
>
> What we have at peerpong is a very simple boot.sh sitting on the
> image. You still have to log in, run boot.sh ->  this registers the
> node on the chef server. then go on the server UI, apply roles to the
> new node.
>
> I guess it would not be difficult to apply roles for the new machine
> from boot.sh using knife though. Command line would look like
> ./boot.sh [database|app||...]
>
> Might spend some time on that tomorrow actually :)
>
> --Gilles
>
> On Sun, May 16, 2010 at 6:41 PM, Cameron 
> Pope<
>  >
>   wrote:
>    
> > For what it's worth, I've been using PoolParty
> > (http://auser.github.com/poolparty/) to define multi-machine deployments on
> > EC2 and I've found that it works pretty well. There have been a few quirks
> > but it's under active development.
> >
> > Cheers!
> > -Cameorn
> >
> > Cameron Pope
> >
> >  
> >
> >
> >
> >
> >      
>