Puppet users - Aug 2011 - How to query other classes/defines configurations in the same node?

I have a huge library of custom classes/defines for all my servers and
I consider myself fluent in puppet DSL. Still there are some
situations I still find hard to handle and was wondering if someone
can help me find a better way to handle these:

Case 1: One define/class requires knowledge of another class in the
same node

  Suppose we have a amanda_dir define and a zabbix_agent class. The
amanda_dir
  defines a amanda directory for backup on the node and the
zabbix_agent class
  installs and configures basic zabbix monitoring on the node.

  My current configuration allows me to add a zabbix_server variable
to the amanda_dir
  define. If this variable is set then the backup is set in a way so
it reports to the zabbix
  server if the backup was successful or send alarm is an error
occurred.

  define amanda_dir($path, $zabbix_server = "") {
      ....
  }

  class zabbix_agent($server) {
      ....
  }

  node my_node {
    class  { "zabbix_agent": server => "10.13.1.1" }
    amanda_dir { "/etc":
       zabbix_enable => true,
       zabbix_server => "10.13.1.1"
    }
  }

  As you see there is duplicated information in the node definition
(the zabbix server IP).
  Ideally the amanda_dir define should be able to determine if the
zabbix_agent class is included in
  the same node and be able to find the zabbix server value and
configure itself to send backup status
  reports to the configured zabbix server without need to replicate
server information or relying on node
  scoped variables or exported resources:

  define amanda_dir($path) {
     if defined(Class["zabbix_agent"])  {
       $zabbix_server = Class["zabbix_agent::server"]
        .....
     }
   }

  this way the amanda_dir define does not require the additional
zabbix_server to be set explicitly.

  This is the simplest example I have but I have several other
situations were a define/class can
  greatly benefit from the ability to query information from another
included classes in the same node.
  The most common one is installing certain versions of packages
depending on previous installed
   versions of libraries (e.g. Nginx+Passenger package version depends
on Ruby package version).

Case 2: Two or more defines/classes need to modify the same file:

  I have a openvpn_tunnel define that creates a openvpn server
configuration file at
  /etc/openvpn/$name-server.conf. Also I have a openvpn_route define
that allows me
  to configure routes to be pushed to the openvpn clients when they
connect. The problem
  is that the routes information has to be added to the corresponding
tunnel configuration file
  /etc/openvpn/$name-server.conf.

  So the situation is that every time puppet runs in the node the
openvpn_tunnel define creates
  the server configuration file and the openvpn_route define modifies
this file adding the desired
  route. The next time puppet runs the openvpn_tunnel define will see
the conf file has been
  modified so it recreates it (deleting the added routes) and then the
openvpn_tunnel define will
  see the routes removed and add them again. This creates an infinite
loop that never ends.

 Case 3: File transfer between nodes

   Some cases certain generated files in one node need to be available
on other nodes (e.g. ssh keys,
   ssl certs, etc.). Current solution is pre-generate these and then
distribute them via the puppet fileserver
   but sometimes I require the file to be generated dynamically at the
specific node because it requires local
   information that may change with time (e.g. DHCP assigned IPs).

-- 
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.

Case 1:

I have gone to a model where I have a manifests/conf directory akin
to /etc which contains the configuration for each service and
application.  In your case I would have a zabbix_conf class and in the
node, I would simple reference the var $zabbix_conf::server_ip.

The beauty of this is my clients know to just go to the conf dir to
make config changes.

Case 2:

I run into this one often and simply get around it with a defined
test:

if ! defined File[/foo] { file { ...} }

In some other situations I use virtual resources such as for
application/services users.  I again have a Conf file that defines the
resource and later realize it.

class users {
   @user { ''jboss'': ....}
}

Then in various places I can safely instanciate the resource:

realize users::user[''jboss'']

Case 3:

A) utilize storedconfigs.

B) create a function that is called by the node on each update which
writes to the puppet file server directory.  The dynamic content can
be passed to the function using the template or inline_template
functions.

HTH.

John



-- 
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.

Case 1:  if you keep your data elsewhere you can use extlookup or Hiera to
retrieve it as needed.

Case 2:  you may want to look at concat,  it''s tailor made for this
scenario.
On Aug 3, 2011 6:45 PM, "hsanson" <hsanson@gmail.com>
wrote:
> I have a huge library of custom classes/defines for all my servers and
> I consider myself fluent in puppet DSL. Still there are some
> situations I still find hard to handle and was wondering if someone
> can help me find a better way to handle these:
>
> Case 1: One define/class requires knowledge of another class in the
> same node
>
> Suppose we have a amanda_dir define and a zabbix_agent class. The
> amanda_dir
> defines a amanda directory for backup on the node and the
> zabbix_agent class
> installs and configures basic zabbix monitoring on the node.
>
> My current configuration allows me to add a zabbix_server variable
> to the amanda_dir
> define. If this variable is set then the backup is set in a way so
> it reports to the zabbix
> server if the backup was successful or send alarm is an error
> occurred.
>
> define amanda_dir($path, $zabbix_server = "") {
> ....
> }
>
> class zabbix_agent($server) {
> ....
> }
>
> node my_node {
> class { "zabbix_agent": server => "10.13.1.1" }
> amanda_dir { "/etc":
> zabbix_enable => true,
> zabbix_server => "10.13.1.1"
> }
> }
>
> As you see there is duplicated information in the node definition
> (the zabbix server IP).
> Ideally the amanda_dir define should be able to determine if the
> zabbix_agent class is included in
> the same node and be able to find the zabbix server value and
> configure itself to send backup status
> reports to the configured zabbix server without need to replicate
> server information or relying on node
> scoped variables or exported resources:
>
> define amanda_dir($path) {
> if defined(Class["zabbix_agent"]) {
> $zabbix_server = Class["zabbix_agent::server"]
> .....
> }
> }
>
> this way the amanda_dir define does not require the additional
> zabbix_server to be set explicitly.
>
> This is the simplest example I have but I have several other
> situations were a define/class can
> greatly benefit from the ability to query information from another
> included classes in the same node.
> The most common one is installing certain versions of packages
> depending on previous installed
> versions of libraries (e.g. Nginx+Passenger package version depends
> on Ruby package version).
>
> Case 2: Two or more defines/classes need to modify the same file:
>
> I have a openvpn_tunnel define that creates a openvpn server
> configuration file at
> /etc/openvpn/$name-server.conf. Also I have a openvpn_route define
> that allows me
> to configure routes to be pushed to the openvpn clients when they
> connect. The problem
> is that the routes information has to be added to the corresponding
> tunnel configuration file
> /etc/openvpn/$name-server.conf.
>
> So the situation is that every time puppet runs in the node the
> openvpn_tunnel define creates
> the server configuration file and the openvpn_route define modifies
> this file adding the desired
> route. The next time puppet runs the openvpn_tunnel define will see
> the conf file has been
> modified so it recreates it (deleting the added routes) and then the
> openvpn_tunnel define will
> see the routes removed and add them again. This creates an infinite
> loop that never ends.
>
> Case 3: File transfer between nodes
>
> Some cases certain generated files in one node need to be available
> on other nodes (e.g. ssh keys,
> ssl certs, etc.). Current solution is pre-generate these and then
> distribute them via the puppet fileserver
> but sometimes I require the file to be generated dynamically at the
> specific node because it requires local
> information that may change with time (e.g. DHCP assigned IPs).
>
> --
> 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.
>
-- 
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.