On Sat, Aug 8, 2009 at 3:20 AM, Steven VanDevender
<stevev@uoregon.edu>wrote:
>
> However, what I''d really prefer to do is use the external node
> classifier only to supply the setting of the Puppet "environment"
> variable, and leave all of the class membership details in our existing
> nodes.pp (included from site.pp). The documentation suggests this
> should probably work, but also says things like "If you specify any
> nodes in your site.pp, you must specify all nodes in site.pp, even those
> that are looked up via the external nodes script." What exactly does
> that mean? We currently have a number of Puppet clients that just use a
> default template without having an explicit node definition; would those
> have to have explicit entries in our nodes.pp rather than falling into
> the "node default"?
When we migrated to external nodes (which was about 2 years ago), we decided
not to have node data at all, only classes.
you can then choose a default class in your external nodes why apply to all
of these hosts.
additionally, we provide as a parameter the environment name, which in turn
configures puppet itself (using a template) to use the correct environment.
>
>
> We''re also starting to use Cobbler for automated installation, and
I''ve
> looked at the configuration management hooks in Cobbler that can be used
> with an external node classifier in Puppet. These are somewhat
> appealing, but again I''d rather not have to migrate all the class
> membership information into Cobbler, and only use the "kickstart
> metadata" field in a host profile to specify whether a host is in the
> development environment. Also, since using Cobbler as an external node
> classifier requires fetching URLs out of the Cobbler web interface to
> get the host data, I am nervous about causing problems for Puppet if
> Cobbler happens to be down.
well, there a lot of solutions for that, you could simply cache your yaml
files on the puppetmaster (as i assume they don''t change so often),
then
tell cobbler to resync the yaml files.
we decided back than that we store our external nodes data in DNS TXT
records... (as we realized that was one of the HA db services that we have
everywhere).
>
>
> If anyone else is using Cobbler for their Puppet external node
> classifier, do you just have to keep Cobbler up all the time, or do you
> have a way to keep the external node classifier working even when
> Cobbler isn''t up at the time?
we ended up having our own full blown web interface which takes care for the
initial provisioning (kickstart, jumpstart preseed etc) and also provide
the external node data to puppet.
the main reasons for not using cobbler is that:
1. it didnt support non kickstart installations
2. we wanted only one kickstart template, it made very little sense to have
multiple kickstart templates with puppet, in our setup, each machine is
"blank" (minimal os installation) and then puppet turns it to the
right
setup based on the external node data, so in that sense,cobbler was an
overkill.
3. we have multiple locations / subnets / and other interfaces, that made
cobbler not so scalable.
4. we wanted to have only one place that a low level technication can go to
and install a server without really understanding any of the infrastrucutre.
hope this helps a bit,
Ohad
>
>
>
> >
>
--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups
"Puppet Users" group.
To post to this group, send email to puppet-users@googlegroups.com
To unsubscribe from this group, send email to
puppet-users+unsubscribe@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/puppet-users?hl=en
-~----------~----~----~----~------~----~------~--~---