Hey Guys,
Let me try to step through this one step at a time. I recommend
downloading one of our VMs that we host for testing puppet (I'm using the
VM we use for training located here -->
http://downloads.puppetlabs.com/training/puppet-vmware.zip but there's also
one for VirtualBox too -->
http://downloads.puppetlabs.com/training/puppet-vbox.zip) to test this
setup. If you do that, make sure DNS and networking are working (So, setup
host entries on your laptop for each VM, setup the hostname and FQDN ON
EACH VM, and make sure each VM can ping the VM Master). I'm using three
VMs (a master and two agents): nodemaster.puppetlabs.vm,
node1.puppetlabs.vm, and node2.puppetlabs.vm. Once DNS resolves and is
working, setup Puppet on the master FIRST, and then setup the remaining
nodes as agents. Again, I recommend using the VMs and using Puppet
Enterprise because it makes it MUCH EASIER to setup Puppet - just use the
puppet-enterprise-installer script (once you get the process of using a
single cert and multiple node_names DOWN, THEN you can start on some actual
machines).
Now we can start customizing Puppet. There are a couple of key steps that
need to be taken since we're shipping around a single cert for all machines:
1. When you stand up a PE Agent, it's going to contact the master and setup
a Certificate Signing Request. Because of this, go to the master node's
$ssldir (which, on Puppet Enterprise, is /etc/puppetlabs/puppet/ssl. If
you want to find the $ssldir on YOUR master, just run `puppet config print
ssldir` and it will print it out for you) and remove the CSRs that are in
the ca/requests/ directory (they should be named after your nodes).
2. Next, lets generate the single certificate that will be shopped around.
You will need to use the same 'certname' on every node. This is the name
of the node as SSL is aware (in my case I'm using 'macs.puppetlabs.vm').
Do this by doing `puppet cert generate macs.puppetlabs.vm` on your MASTER
machine. The master will generate the certs and your $ssldir will look
something like this:
ssl
|-- ca
| |-- ca_crl.pem
| |-- ca_crt.pem
| |-- ca_key.pem
| |-- ca_pub.pem
| |-- inventory.txt
| |-- private
| | `-- ca.pass
| |-- requests
| |-- serial
| `-- signed
| |-- macs.puppetlabs.vm.pem
| |-- nodemaster.puppetlabs.vm.pem
| |-- pe-internal-broker.pem
| |-- pe-internal-dashboard.pem
| |-- pe-internal-mcollective-servers.pem
| |-- pe-internal-peadmin-mcollective-client.pem
| `-- pe-internal-puppet-console-mcollective-client.pem
|-- certificate_requests
|-- certs
| |-- ca.pem
| |-- macs.puppetlabs.vm.pem
| |-- nodemaster.puppetlabs.vm.pem
| |-- pe-internal-broker.pem
| |-- pe-internal-mcollective-servers.pem
| |-- pe-internal-peadmin-mcollective-client.pem
| `-- pe-internal-puppet-console-mcollective-client.pem
|-- crl.pem
|-- private
|-- private_keys
| |-- macs.puppetlabs.vm.pem
| |-- nodemaster.puppetlabs.vm.pem
| |-- pe-internal-broker.pem
| |-- pe-internal-mcollective-servers.pem
| |-- pe-internal-peadmin-mcollective-client.pem
| `-- pe-internal-puppet-console-mcollective-client.pem
`-- public_keys
|-- foo.puppetlabs.vm.pem
|-- macs.puppetlabs.vm.pem
|-- nodemaster.puppetlabs.vm.pem
|-- pe-internal-broker.pem
|-- pe-internal-mcollective-servers.pem
|-- pe-internal-peadmin-mcollective-client.pem
`-- pe-internal-puppet-console-mcollective-client.pem
3. There are three files you need to collect on your master and ship
around to all of your nodes. They are ALL in the $ssldir, so these paths
are relative to THAT directory. The three files are:
- $ssldir/private_keys/macs.puppetlabs.vm.pem on the master -> gets
copied to the agent's $ssldir/private_keys directory
- $ssldir/public_keys/macs.puppetlabs.vm.pem on the master -> gets
copied to the agent's $ssldir/public_keys directory
- $ssldir/ca/signed/macs.puppetlabs.vm.pem on the master -> gets
copied to the agent's $ssldir/certs directory
4. You need to modify /etc/puppetlabs/puppet/auth.conf on the master so
every node can access the find action on the catalog REST endpoint for the
master (i.e. The Mac nodes can get their catalog). Look for this stanza in
auth.conf:
# allow nodes to retrieve their own catalog (ie their configuration)
path ~ ^/catalog/([^/]+)$
method find
allow $1
Add the following line under 'allow $1': (Remember that I'm using the
certname of macs.puppetlabs.vm - you would substitute the certname you will
use in your infrastructure)
- allow macs.puppetlabs.vm
5. Now, we need to modify /etc/puppetlabs/puppet/puppet.conf on the NODES
themselves. There are two changes that need to be made: the certname and
the nodename. Remember that we need to decouple the name that SSL uses to
identify the node (macs.puppetlabs.vm) with the name that PUPPET uses to
CLASSIFY the node (node1.puppetlabs.vm and node2.puppetlabs.vm in our
case). The first line you need to change in puppet.conf is the certname
configuration item, set that to the following on ALL of your nodes:
'certname = macs.puppetlabs.vm'. If you're using Puppet Enterprise, that
item should already be in the [agent] stanza, so you'll need to change it.
The next configuration item could either go in the [agent] or [main]
stanza, and that's the node_name_fact OR the node_name_value item. NOTE:
THESE ITEMS ARE MUTUALLY EXCLUSIVE - YOU CAN ONLY USE ONE OR THE OTHER.
For more information, see -->
http://docs.puppetlabs.com/references/stable/configuration.html#nodenamefact
The node_name_fact allows you to set the nodename based on a Facter
fact.
In my case I'm going to set 'node_name_fact = fqdn' on all of my nodes
because I want the nodename to match the FQDN on the machine. If you're
managing Macs, however, you might want to use something relatively static
like a serial number (if you use something like the hostname, it will
change any time someone renames their machine in the sharing pane). You
could also do 'node_name_value = thisnode' and Puppet would classify the
node based on the name of 'thisnode', but you would need to have that item
be unique for every one of your nodes. This is a change you will need to
evaluate in your own environment.
6. Once you've generated the cert, put it in the appropriate directories
on the client, changed auth.conf, set puppet.conf on the agents, and
everything is installed, you should be able to run `puppet agent -t` and
watch things work! Note that if you're using the console, you will see the
node records listed based on the nodename.
Let me know if you have any other questions on this process :)
On Thu, Apr 12, 2012 at 7:58 AM, Sean McGrath <seanc.mcgr...@gmail.com>wrote:
> Gary,
>
> Thanks very much for getting back to me on this. While I have been
> researching Puppet for our Mac fleet I have used a lot of the very
> useful information you have published about this so thank you very
> much for that.
>
> In regards your response. Firstly, apologies for my lack of knowledge
> in this matter and any help that can be provided will be very much
> appreciated.
>
> I am interested in #3 and am having looking at it now but the
> following are some of the things that have crossed my mind
>
> The private cert to distribute to the client nodes, which one from the
> master is it? There are a few there and I cannot tell which one it is
> and where does it go on the client?
>
> Is there any other configuration that is needed on the master?
>
> In the clients /etc/puppet/puppet.conf file would the following
> configuration be correct for this approach.
>
> [puppetd]
> node_name_fact = hostname
> node_name_value = certname # i.e. the one from the server
>
> Is this correct?
>
> Thats it for now, I'm going to stay playing around with this and see
> where I can get but any help and guidance that anyone can provide will
> be very much appreciated.
>
> Regards
>
> Sean
>
> On Apr 11, 5:32 pm, Gary Larizza <g...@puppetlabs.com> wrote:
> > Hey Sean,
> >
> > First - congrats on wrangling your Macs with Puppet! Next, I understand
> and
> > have shared your pain regarding timely imaging of workstations and Puppet
> > cert-wrangling. Generally, I've seen folks do one of a couple of things:
> >
> > 1. Autosign
> > 2. Utilize a CGI script to sign/revoke certs on the master (which can
> > largely be replaced through the use of the `puppet cert` face)
> > 3. Use the same private key everywhere and change the individual
> > node_name
> >
> > Numbers 1 and 2 are largely process around signing individual certs for
> > every node. You COULD even backup the $ssldir on your clients, image the
> > machine, install puppet, restore the $ssldir, and then run Puppet again
> and
> > Puppet will work fine for your clients.
> >
> > Number 3 is a bit different. With #3, you would have the SAME private
> cert
> > for EVERY node in your infrastructure. Because of this, the certname
> must
> > be THE SAME for every node. When you do this, however, Puppet treats
> every
> > node as if it were the SAME node - so you need a way to de-couple the
> name
> > of the node as Puppet knows it with the name of the node as the
> Certificate
> > knows it. The solution is the 'node_name_fact' and 'node_name_value'
> > configuration item in puppet.conf -->
> http://docs.puppetlabs.com/references/stable/configuration.html#noden...
> > You would essentially ship the private cert around to EVERY node, set
> > the
> > node_name_{fact,value} in puppet.conf, and then Puppet would treat each
> > machine as a separate node (even though the certificate is the same
> > everywhere). Obviously there are security implications for this, but
> some
> > people prefer it to Autosigning.
> >
> > Hopefully, this should help you on your way.
> >
> > On Wed, Apr 11, 2012 at 8:31 AM, Sean McGrath <seanc.mcgr...@gmail.com
> >wrote:
> >
> >
> >
> >
> >
> >
> >
> >
> >
> > > Firstly my apologies for posting this if it has been answered
> > > elsewhere and I missed it while looking.
> >
> > > I'm starting to look at using Puppet to manage our fleet of Mac's
> > > running OS X in our lab environment and I'm quite impressed with it
> > > from my testing so far.
> >
> > > I have tested the functionality of the autosign.conf file with the
> > > hostnames of the trusted clients in it.
> >
> > > However, if I re-image one of the Mac's as we occasionally do that
> > > destroys the client certificate that it uses for the puppetca request.
> > > Thus the puppet master see's a request with a different certificate
> > > from a node with a hostname that has had its trust relationship
> > > established with a different certificate.
> >
> > > This is probably a noob question but I haven't been able to figure it
> > > out. How do I get around this in an automated manner. I don't want to
> > > have to revoke certificates each time I re-image a Mac so they can be
> > > re-trusted by the puppet master. Is there something like a root
> > > certificate I could build into the image to establish the trust
> > > relationship easily and securely each time a Mac is re-imaged?
> >
> > > many thanks
> >
> > > Sean
> >
> > > --
> > > 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+unsubscr...@googlegroups.com.
> > > For more options, visit this group at
> > >http://groups.google.com/group/puppet-users?hl=en.
> >
> > --
> >
> > Gary Larizza
> > Professional Services Engineer
> > Puppet Labs
>
> --
> 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+unsubscr...@googlegroups.com.
> For more options, visit this group at
> http://groups.google.com/group/puppet-users?hl=en.
>
>
--
Gary Larizza
Professional Services Engineer
Puppet Labs
--
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+unsubscr...@googlegroups.com.
For more options, visit this group at
http://groups.google.com/group/puppet-users?hl=en.
