Puppet users - Apr 2007 - dependency and communication between defined classes

Hi,

 i wanted to know how you handle case when classes or define need to 
communicate between them. For exemple i got an ftpd define and a 
apachevhost define. Both need to know the path where the vhost is set 
and this path is defined by the ftpuser home''s directory. How can i ask
information from other define or other classes ?  we allready seen that 
tag are not reliable as they depend on the order the parser read them.

  For exemple how to say that a define requires another define  as a 
requirement ( i think of the ftpuser define that creates a ftp user 
account and the ftpdaemon define that is used to configure the ftp 
daemon and the define that create the apache vhost ,it need the ftpuser 
to be created before and the ftpuser define  need that the ftpdaemon 
exist before ...).

  i also tried to make a define inside another define but i cannot. I 
have seen the syntax xx::yy to acces some data but when is this usable ? 
What data can be read between part of the recipes ?

  I see my recipe getting more complex and i am a little lost on how 
they can interact with each others. Any ideas/hint on this side ?

-- 
Cordialement,
Ghislain
 



_______________________________________________
Puppet-users mailing list
Puppet-users@madstop.com
https://mail.madstop.com/mailman/listinfo/puppet-users

On Apr 18, 2007, at 3:45 PM, ADNET Ghislain wrote:
> Hi,
>
> i wanted to know how you handle case when classes or define need to  
> communicate between them. For exemple i got an ftpd define and a  
> apachevhost define. Both need to know the path where the vhost is  
> set and this path is defined by the ftpuser home''s directory. How
> can i ask information from other define or other classes ?  we  
> allready seen that tag are not reliable as they depend on the order  
> the parser read them.
>
>  For exemple how to say that a define requires another define  as a  
> requirement ( i think of the ftpuser define that creates a ftp user  
> account and the ftpdaemon define that is used to configure the ftp  
> daemon and the define that create the apache vhost ,it need the  
> ftpuser to be created before and the ftpuser define  need that the  
> ftpdaemon exist before ...).
>
>  i also tried to make a define inside another define but i cannot.  
> I have seen the syntax xx::yy to acces some data but when is this  
> usable ? What data can be read between part of the recipes ?
Right now, no such thing exists.  Any recommendations on how to do  
this would be great.

  --
  A cult is a religion with no political power.   -- Tom Wolfe
  ---------------------------------------------------------------------
  Luke Kanies | http://reductivelabs.com | http://madstop.com

On Apr 18, 2007, at 2:42 PM, Luke Kanies wrote:
> On Apr 18, 2007, at 3:45 PM, ADNET Ghislain wrote:
>
>> Hi,
>>
>> i wanted to know how you handle case when classes or define need to
>> communicate between them. For exemple i got an ftpd define and a
>> apachevhost define. Both need to know the path where the vhost is
>> set and this path is defined by the ftpuser home''s directory.
How
>> can i ask information from other define or other classes ?  we
>> allready seen that tag are not reliable as they depend on the order
>> the parser read them.
>>
>>  For exemple how to say that a define requires another define  as a
>> requirement ( i think of the ftpuser define that creates a ftp user
>> account and the ftpdaemon define that is used to configure the ftp
>> daemon and the define that create the apache vhost ,it need the
>> ftpuser to be created before and the ftpuser define  need that the
>> ftpdaemon exist before ...).
>>
>>  i also tried to make a define inside another define but i cannot.
>> I have seen the syntax xx::yy to acces some data but when is this
>> usable ? What data can be read between part of the recipes ?
>
> Right now, no such thing exists.  Any recommendations on how to do
> this would be great.
Doesn''t load order allow this?  For example:

class conglomeration {
     define { ... }
     include class_with_definition_that_uses_variables_from_above_define
}

-Blake

On Apr 18, 2007, at 6:21 PM, Blake Barnett wrote:
>
> Doesn''t load order allow this?  For example:
>
> class conglomeration {
>      define { ... }
>      include  
> class_with_definition_that_uses_variables_from_above_define
> }
Nope; variables are available to scopes below, but not above or at  
the same level.  In other words, variables set in the conglomeration  
class are available in both the define and class, but that''s it.

  --
  What is the sound of Perl?  Is it not the sound of a wall that
  people have stopped banging their heads against?
          --Larry Wall in <1992Aug26.184221.29627@netlabs.com>
  ---------------------------------------------------------------------
  Luke Kanies | http://reductivelabs.com | http://madstop.com

On Apr 18, 2007, at 3:45 PM, ADNET Ghislain wrote:
> Hi,
>
> i wanted to know how you handle case when classes or define need to  
> communicate between them. For exemple i got an ftpd define and a  
> apachevhost define. Both need to know the path where the vhost is  
> set and this path is defined by the ftpuser home''s directory. How
> can i ask information from other define or other classes ?  we  
> allready seen that tag are not reliable as they depend on the order  
> the parser read them.
>
>  For exemple how to say that a define requires another define  as a  
> requirement ( i think of the ftpuser define that creates a ftp user  
> account and the ftpdaemon define that is used to configure the ftp  
> daemon and the define that create the apache vhost ,it need the  
> ftpuser to be created before and the ftpuser define  need that the  
> ftpdaemon exist before ...).
>
>  i also tried to make a define inside another define but i cannot.  
> I have seen the syntax xx::yy to acces some data but when is this  
> usable ? What data can be read between part of the recipes ?
After berating myself a bit for having my own language that doesn''t  
do what people want, I asked myself how hard this would be, and I  
concluded, not that hard.  After thinking about it some to make sure  
it doesn''t open up any weird holes anywhere, I''ve added the  
functionality.  The following code now works as you''d hope (in svn,  
of course):

class one {
     $var = value
}

class one::two {
     $other = more
}

class three {
     include one, one::two
     $three = $one::two::other
     exec { "/bin/echo one: $one::var, two: $one::two::other, three:  
$three": }
}
include three

You cannot (and likely never will be able to) assign to variables in  
other classes, and you *must* evaluate the class before you try to  
retrieve its variables (thus the ''include'' in the
''three'' class).

This is in changeset 2393:

https://reductivelabs.com/trac/puppet/changeset/2393

Comments?

  --
  Always behave like a duck - keep calm and unruffled on the surface but
  paddle like the devil underneath.   -- Jacob Braude
  ---------------------------------------------------------------------
  Luke Kanies | http://reductivelabs.com | http://madstop.com

On Apr 19, 2007, at 1:40 PM, Luke Kanies wrote:
>
> After berating myself a bit for having my own language that
doesn''t
> do what people want, I asked myself how hard this would be, and I
> concluded, not that hard.  After thinking about it some to make sure
> it doesn''t open up any weird holes anywhere, I''ve added
the
> functionality.  The following code now works as you''d hope (in
svn,
> of course):
>
> class one {
>      $var = value
> }
>
> class one::two {
>      $other = more
> }
>
> class three {
>      include one, one::two
>      $three = $one::two::other
>      exec { "/bin/echo one: $one::var, two: $one::two::other, three:
> $three": }
> }
> include three
You can also use ''$::var'' to refer to variables in the
top-level scope.

I realize now that people are going to want references now, because  
you''re going to want to refer to variables set within a node config,  
and that''s going to be impossible because this code won''t
work:

   $var = ${$hostname}::variable

Hmm.  Would that be reasonable to add?  Or should we instead make the  
class ''node'' semi-magical, so that you could say:

   $var = $node::variable

Maybe I should just add a function like ''nodevar'':

   $var = nodevar(variable)

That would correctly look through the tree of node scopes, so that  
nodes could override values set in parent nodes.

Ideas?

  --
  To be positive: To be mistaken at the top of one''s voice.
          -- Ambrose Bierce
  ---------------------------------------------------------------------
  Luke Kanies | http://reductivelabs.com | http://madstop.com

On Apr 19, 2007, at 11:53 AM, Luke Kanies wrote:
> On Apr 19, 2007, at 1:40 PM, Luke Kanies wrote:
>>
>> After berating myself a bit for having my own language that
doesn''t
>> do what people want, I asked myself how hard this would be, and I
>> concluded, not that hard.  After thinking about it some to make sure
>> it doesn''t open up any weird holes anywhere, I''ve
added the
>> functionality.  The following code now works as you''d hope (in
svn,
>> of course):
>>
>> class one {
>>      $var = value
>> }
>>
>> class one::two {
>>      $other = more
>> }
>>
>> class three {
>>      include one, one::two
>>      $three = $one::two::other
>>      exec { "/bin/echo one: $one::var, two: $one::two::other,
three:
>> $three": }
>> }
>> include three
>
> You can also use ''$::var'' to refer to variables in the
top-level
> scope.
>
> I realize now that people are going to want references now, because
> you''re going to want to refer to variables set within a node
config,
> and that''s going to be impossible because this code won''t
work:
>
>    $var = ${$hostname}::variable
>
> Hmm.  Would that be reasonable to add?  Or should we instead make the
> class ''node'' semi-magical, so that you could say:
>
>    $var = $node::variable
>
> Maybe I should just add a function like ''nodevar'':
>
>    $var = nodevar(variable)
>
> That would correctly look through the tree of node scopes, so that
> nodes could override values set in parent nodes.
I like this last one.

-Blake

>> Maybe I should just add a function like ''nodevar'':
>>
>>    $var = nodevar(variable)
>>
>> That would correctly look through the tree of node scopes, so that
>> nodes could override values set in parent nodes.
>>     
>
> I like this last one.
>
> -Blake
same here, the others seems...weird in syntax to me :)


-- 
Cordialement,
Ghislain
 



_______________________________________________
Puppet-users mailing list
Puppet-users@madstop.com
https://mail.madstop.com/mailman/listinfo/puppet-users

> This is in changeset 2393:
>
> https://reductivelabs.com/trac/puppet/changeset/2393
>
> Comments?
>
>   
this is a great thing, let me some time to see how i can work my recipe 
with this feature and i will tell you if i have a comment :)

at first i wondered if one::two::$other woudl be easier to read or 
${one::two::other} ..humm not sure yet...


-- 
Cordialement,
Ghislain


_______________________________________________
Puppet-users mailing list
Puppet-users@madstop.com
https://mail.madstop.com/mailman/listinfo/puppet-users

On Apr 19, 2007, at 5:59 PM, ADNET Ghislain wrote:
> this is a great thing, let me some time to see how i can work my  
> recipe with this feature and i will tell you if i have a comment :)
>
> at first i wondered if one::two::$other woudl be easier to read or $ 
> {one::two::other} ..humm not sure yet...
The first would be much harder to lex -- you wouldn''t know you had a  
variable until you got to the $.  This version was pretty easy to  
lex, and it''s pretty easy to understand, I think.

  --
  It''s impossible to foresee the consequences of being clever.
              -- Christopher Strachey
  ---------------------------------------------------------------------
  Luke Kanies | http://reductivelabs.com | http://madstop.com

would it be also possible to a way to make user defined
''define'' require
something in the definition and not in the calling of it ?

i mean you can do

define test($testme){
    file { test: }
}

test { montest:
          testme => ''hello'',
          require   =>   File[''debug'']
}

but if i want that my define is allways requiring something i cannot (or 
do not know) how to say :

define test($testme){
    file { test: }
    require    =>   File[''debug'']
}

so that the user do not have to know the internal of the class and put 
requires on all calls of it as the define take cares of it... same for 
before, refresh etc...


-- 
Cordialement,
Ghislain


_______________________________________________
Puppet-users mailing list
Puppet-users@madstop.com
https://mail.madstop.com/mailman/listinfo/puppet-users

ADNET Ghislain schrieb:
> would it be also possible to a way to make user defined
''define''
> require something in the definition and not in the calling of it ?
>
> i mean you can do
>
> define test($testme){
>    file { test: }
> }
>
> test { montest:
>          testme => ''hello'',
>          require   =>   File[''debug'']
> }
>
> but if i want that my define is allways requiring something i cannot 
> (or do not know) how to say :
>
> define test($testme){
>    file { test: }
>    require    =>   File[''debug'']
> }
>
> so that the user do not have to know the internal of the class and put 
> requires on all calls of it as the define take cares of it... same for 
> before, refresh etc...
>
>
> ------------------------------------------------------------------------
>
> _______________________________________________
> Puppet-users mailing list
> Puppet-users@madstop.com
> https://mail.madstop.com/mailman/listinfo/puppet-users
>   
When I understand you correctly you want this, as defines implicitly 
"require" all described objects:

define test($testme, $pathvar){
   file { test: }
  
   file {''debugfile'':
         path => "$pathvar"
}

where $pathvar is just introduced to not break the convention to not use 
anything but variables in defines.
hope i understood your question correctly..

this is not really the same thing as here you create the 
File[''debugfile''] in your define, what i want is that my
define be run
AFTER the File[''debugfile''] is processed but this debugfile is
NOT
created by the define but comes from another part of the recipe.

 In your solution File[''debugfile''] cannot exist before or you
have a
duplicate definition issue.
> When I understand you correctly you want this, as defines implicitly 
> "require" all described objects:
>
> define test($testme, $pathvar){
>    file { test: }
>   
>    file {''debugfile'':
>          path => "$pathvar"
> }
>
> where $pathvar is just introduced to not break the convention to not use 
> anything but variables in defines.
> hope i understood your question correctly..
>
>   

-- 
Cordialement,
Ghislain ADNET.
AQUEOS.

Attention !  Toute demande de support ou commande de domaine par email 
sera refusée, pour cela utilisez https://support.aqueos.net.  

Pour tout contact nos coordonnées :  
http://www.aqueos.com/aqueos-services-informatiques-societe.php
Fax: 01.72.70.32.66
 




_______________________________________________
Puppet-users mailing list
Puppet-users@madstop.com
https://mail.madstop.com/mailman/listinfo/puppet-users

You could create the debugfile as a virual resource and create a fixed 
order in your definition of "test" by using requires. I guess this
could
get quite ugly in the long run but depending on your actual case I think 
this could be a temporary solution..

ADNET Ghislain schrieb:
> this is not really the same thing as here you create the 
> File[''debugfile''] in your define, what i want is that my
define be run
> AFTER the File[''debugfile''] is processed but this
debugfile is NOT
> created by the define but comes from another part of the recipe.
>
> In your solution File[''debugfile''] cannot exist before or
you have a
> duplicate definition issue.
>> When I understand you correctly you want this, as defines implicitly 
>> "require" all described objects:
>>
>> define test($testme, $pathvar){
>>    file { test: }
>>      file {''debugfile'':
>>          path => "$pathvar"
>> }
>>
>> where $pathvar is just introduced to not break the convention to not 
>> use anything but variables in defines.
>> hope i understood your question correctly..
>>
>>   
>
>
> ------------------------------------------------------------------------
>
> _______________________________________________
> Puppet-users mailing list
> Puppet-users@madstop.com
> https://mail.madstop.com/mailman/listinfo/puppet-users
>

On Apr 19, 2007, at 2:21 PM, Blake Barnett wrote:
>> Maybe I should just add a function like ''nodevar'':
>>
>>    $var = nodevar(variable)
>>
>> That would correctly look through the tree of node scopes, so that
>> nodes could override values set in parent nodes.
>
> I like this last one.
I looked into creating this last night, and it turns out that it''s  
going to be quite a bit of work because of the order in which  
statements are interpreted and the fact that nodes evaluate all of  
the other classes.  In particular, look at this configuration:

class testing {
   $val = nodevar("myval")
}

node basenode {
   $myval = something
   include testing
}

node mynode inherits basenode {
   $myval = somethingelse
}

In this case, the basenode gets evaluated before mynode, and part of  
evaluating the ''basenode'' involves evaluating the
''testing'' class,
which means that it''s evaluated before mynode is, which means you  
don''t get the values you want.

I''ve been trying to think of simple ways to resolve this, and I
don''t
think there are any.  I think the only real resolution is to rethink  
ordering as a whole within the parsing process, and likely  
introducing the ability to declare that a given class requires  
another, so that we can always guarantee that the required class is  
evaluated first and thus its variables are set in time for the second.

For now, we''re just going to be stuck with no nodevar.

  --
  The great tragedy of Science - the slaying of a beautiful  
hypothesis by
  an ugly fact.   --Thomas H. Huxley
  ---------------------------------------------------------------------
  Luke Kanies | http://reductivelabs.com | http://madstop.com

On Apr 20, 2007, at 2:34 AM, ADNET Ghislain wrote:
>
> define test($testme){
>    file { test: }
>    require    =>   File[''debug'']
> }
>
> so that the user do not have to know the internal of the class and  
> put requires on all calls of it as the define take cares of it...  
> same for before, refresh etc...
You can just add the require directly:

define test($testme) {
   file { test: require => File[''debug''] }
}

But this will make it so that people can''t specify their own  
''require'' values.

There''s no way around that problem right now.

  --
  Brand''s Asymmetry:
     The past can only be known, not changed. The future can only be
     changed, not known.
  ---------------------------------------------------------------------
  Luke Kanies | http://reductivelabs.com | http://madstop.com

> You can just add the require directly:
>
> define test($testme) {
>    file { test: require => File[''debug''] }
> }
>   
yes but that can work for this exemple but if there is 47 objects in the 
define  it make it quite ugly. Also if i define any variable or any 
include in the define they get included even as the require is not 
fullfilled:

define test($testme) {
    include testclass
   file { test: require => File[''debug''] }
}


in my exemple the include would not be "included" as the global
require
would prevent the whole define to run before the require. In your way it 
would be included and then the define would then trigger the require and 
be sceduled later (if i am not mistaken).

this become ordering mayhem !! ;)

-- 
Cordialement,
Ghislain


_______________________________________________
Puppet-users mailing list
Puppet-users@madstop.com
https://mail.madstop.com/mailman/listinfo/puppet-users

On Apr 20, 2007, at 10:17 AM, ADNET Ghislain wrote:
> yes but that can work for this exemple but if there is 47 objects  
> in the define  it make it quite ugly. Also if i define any variable  
> or any include in the define they get included even as the require  
> is not fullfilled:
>
> define test($testme) {
>    include testclass
>   file { test: require => File[''debug''] }
> }
>
>
> in my exemple the include would not be "included" as the global  
> require would prevent the whole define to run before the require.  
> In your way it would be included and then the define would then  
> trigger the require and be sceduled later (if i am not mistaken).
>
> this become ordering mayhem !! ;)
Yeah, I think it''s pretty clear that the language needs more ability  
to specify relationships between groups of resources, but I don''t  
know when I''ll have time to do that.

If this is a big problem for you, I''m available for paid custom  
development to get it done sooner, but otherwise, I can''t make any  
promises about when it will end up on my todo list.

  --
  It is a mistake to think you can solve any major problems just with
  potatoes.                        --Douglas Adams
  ---------------------------------------------------------------------
  Luke Kanies | http://reductivelabs.com | http://madstop.com

Luke Kanies wrote:
> On Apr 20, 2007, at 2:34 AM, ADNET Ghislain wrote:
>> define test($testme){
>>    file { test: }
>>    require    =>   File[''debug'']
>> }
>>
>> so that the user do not have to know the internal of the class and  
>> put requires on all calls of it as the define take cares of it...  
>> same for before, refresh etc...
> 
> You can just add the require directly:
> 
> define test($testme) {
>    file { test: require => File[''debug''] }
> }
> 
> But this will make it so that people can''t specify their own  
> ''require'' values.
> 
> There''s no way around that problem right now.
Sorry for jumping in late, catching up on email.

Maybe what I think is working isn''t, but I''ve been forcing
requires in
my defines and allowing callers to define additional requirements by
using selectors, defaults, and lists.  I believe this works if the
caller passes a singleton or a list.

Take note of $require_real => $require ? { }

# Copy things out of a subversion working copy.
define file_from_cache($source = false,
  $sourcedir = "/Support/vault/cache",
  $destdir = false,
  $owner = 0, $group = 0, $mode = 0640,
  $recurse = true, $ignore = ".svn",
  $backup = false, $require = false )
{
  $source_real = $source ? {
    false => $name,
    default => $source
  }
  $name_real = $destdir ? {
    false => $name,
    default => "$destdir/$name"
  }
  # Require the subversion working copy.
  # This is created in the siteconf-tree class, so include it too.
  $require_real = $require ? {
    false => Svnworkdir["FileCache"],
    default => [ Svnworkdir["FileCache"], $require]
  }
  # We always need the siteconf-tree when copying files from the cache.
  # FIXME: Local only mode?
  include siteconf-tree
  # The actual file to copy from the svn working copy.
  file { $name_real:
    source => "$sourcedir/$source_real",
    owner => $owner,
    group => $group,
    mode => $mode,
    recurse => $recurse,
    ignore => $ignore,
    backup => $backup,
    require => $require_real
  }
}

class siteconf-tree {
  svnworkdir { "FileCache":
    repository => "https://manage.lan/svn/siteconfig/trunk",
    local_name => "cache"
  }
}

-- 
Jeff McCune
The Ohio State University
Department of Mathematics
Systems Manager


_______________________________________________
Puppet-users mailing list
Puppet-users@madstop.com
https://mail.madstop.com/mailman/listinfo/puppet-users

On Apr 26, 2007, at 9:09 AM, Jeff McCune wrote:
> Sorry for jumping in late, catching up on email.
>
> Maybe what I think is working isn''t, but I''ve been
forcing requires in
> my defines and allowing callers to define additional requirements by
> using selectors, defaults, and lists.  I believe this works if the
> caller passes a singleton or a list.
I''d be shocked if it worked if you pass in an array:

define test() {
     exec { "/bin/echo testing $name": subscribe => [Exec[one],  
$subscribe] }
}

exec { one: command => "/bin/echo one", refreshonly => true }
exec { two: command => "/bin/echo two", refreshonly => true }
exec { three: command => "/bin/echo three", refreshonly => true
}
test { yay: subscribe => [Exec[two], Exec[three]] }

Produces:

err: Could not apply complete configuration: Could not retrieve  
dependency ''Exec[two][Exec[three]]'' at
/Users/luke/bin/test2.pp:4

  --
  A diplomat is a man who can convince his wife she''d look stout in a  
fur
  coat.
  ---------------------------------------------------------------------
  Luke Kanies | http://reductivelabs.com | http://madstop.com

Luke Kanies wrote:
> On Apr 26, 2007, at 9:09 AM, Jeff McCune wrote:
>> Sorry for jumping in late, catching up on email.
>>
>> Maybe what I think is working isn''t, but I''ve been
forcing requires in
>> my defines and allowing callers to define additional requirements by
>> using selectors, defaults, and lists.  I believe this works if the
>> caller passes a singleton or a list.
> 
> I''d be shocked if it worked if you pass in an array:
> 
> define test() {
>      exec { "/bin/echo testing $name": subscribe =>
[Exec[one],
> $subscribe] }
> }
> 
> exec { one: command => "/bin/echo one", refreshonly => true
}
> exec { two: command => "/bin/echo two", refreshonly => true
}
> exec { three: command => "/bin/echo three", refreshonly =>
true }
> test { yay: subscribe => [Exec[two], Exec[three]] }
> 
> Produces:
> 
> err: Could not apply complete configuration: Could not retrieve  
> dependency ''Exec[two][Exec[three]]'' at
/Users/luke/bin/test2.pp:4
Yep, you''re right.  Bummer.

-- 
Jeff McCune
The Ohio State University
Department of Mathematics
Systems Manager


_______________________________________________
Puppet-users mailing list
Puppet-users@madstop.com
https://mail.madstop.com/mailman/listinfo/puppet-users