In the file extract.pp

      $extract_zip    = "unzip -o ${src_target}/${name}.${extension} -d ${target}"
      $extract_targz  = "tar --no-same-owner --no-same-permissions --strip-components=${strip_components} -xzf ${src_target}/${name}.${extension} -C ${target}"
      $extract_tarbz2 = "tar --no-same-owner --no-same-permissions --strip-components=${strip_components} -xjf ${src_target}/${name}.${extension} -C ${target}"

These ${target} should be ${extract_dir}

That said: I'm more concerned with how this module isn't consistent with how the parameters should be used. I much prefer having a single $target to extract to because that's conceptually easier to follow.

I have two examples. One which works and another which doesn't. The only difference I can see is that in once case the name of the archive which is downloaded matches the url.

Does work

archive { 'kibana-3.1.2':
    checksum => false,
    ensure => present,
    url    => 'https://download.elasticsearch.org/kibana/kibana/kibana-3.1.2.tar.gz',
    target => '/opt',
  }

Doesn't work ( The file that is downloaded is pguri-master.tar.gz)

  archive { 'pguri-master':
    checksum => false,
    ensure => present,
    url    => 'http://github.com/petere/pguri/archive/master.tar.gz',
    target => '/opt',
  }

Ran into an issue with the following code

 45       $packer_basename = inline_template(
 46         "<%= \"#{@prefix}#{@version}_#{scope['::kernel'].downcase}_#{@arch}.zip\" %>"
 47       )
 48
 49       $packer_url = "${packer::params::base_url}${packer_basename}"
 50
 51       # Download the Packer zip archive to the cache.
 52       archive { $packer_basename :
 53         ensure           => present,
 54         url              => "${packer_url}",
 55         target           => $bin_dir,
 56         follow_redirects => true,
 57         extension        => 'zip',
 58         checksum         => false,
 59         src_target       => $cache_dir,
 60       }

it produced this output, where it appended .zip to the file name.

Notice: /Stage[main]/Packer/Archive[packer_0.7.5_linux_amd64.zip]/Archive::Download[packer_0.7.5_linux_amd64.zip.zip]/Exec[download archive packer_0.7.5_linux_amd64.zip.zip and check sum]/returns: executed successfully
Notice: /Stage[main]/Packer/Archive[packer_0.7.5_linux_amd64.zip]/Archive::Extract[packer_0.7.5_linux_amd64.zip]/Exec[packer_0.7.5_linux_amd64.zip unpack]/returns: executed successfully
Notice: /Stage[main]/Main/Node[xagent.vagrant.vm]/Packer::Plugin[post-processor-vagrant-vmware-ovf]/Archive[packer-post-processor-vagrant-vmware-ovf.linux-amd64]/Archive::Extract[packer-post-processor-vagrant-vmware-ovf.linux-amd64]/Exec[packer-post-processor-vagrant-vmware-ovf.linux-amd64 unpack]/returns: executed successfully

vagrant]# ls /tmp/
packer_0.7.5_linux_amd64.zip.zip

It didn't error but the output was unexpected. Am I missing something?

I get the following failure when I compile via rspec-puppet:

error during compilation: Validation of Exec[download archive kibana-4.3.1-linux-x64.tar.gz and check sum] failed: 'curl  -s -S   -o /usr/src/kibana-4.3.1-linux-x64.tar.gz 'http://myplace/kibana/4.3.1/kibana-4.3.1-x64.tar.gz'' is not qualified and no path was specified. Please qualify the command or specify a path. at spec/fixtures/modules/archive/manifests/download.pp:171

Meanwhile in a627bc3 archive::download now defaults to a fact or global scope variable $::path?

define archive::download (
...
  $path=$::path,
) {

Anyone know what the thinking is here? ping @raphink

The default example does not work with puppet 3.2.

1- Puppet archive has Exec commands defined without qualifier, so the archive resource type will not work if is not fixing this first
2- The dependencies are not being filtered in order. The example

tomcat_mirror = "http://archive.apache.org/dist/tomcat/"
$tomcat_version = "6.0.37"

include tomcat::source

fails with

Error: Scope(Class[Tomcat::Juli]): undefined mandatory attribute: $tomcat_home
Error: Scope(Class[Tomcat::Logging]): undefined mandatory attribute: $tomcat_home
Error: Could not find dependency File[undef] for File[/extras/]

Because the includes of Tomcat::Juli and Tomcat::Logging are performed befre $tomcat_home has been declared in Tomcat::source

It would be nice if the module would support simple tar files (no compression).

My zip file is not on a URL. Does the module support extraction of zip files that are already on the client?

When I try to use the archive module in puppet, i get this log :

Error: Validation of Exec[download archive gitlist.tar.gz and check sum] failed: 'curl -s -S  -L -o /tmp/gitlist.tar.gz https://s3.amazonaws.com/gitlist/gitlist-0.5.0.tar.gz' is not qualified and no path was specified. Please qualify the command or specify a path. at /home/sebastien/backup/modules/archive/manifests/download.pp:146
Wrapped exception:
'curl -s -S  -L -o /tmp/gitlist.tar.gz https://s3.amazonaws.com/gitlist/gitlist-0.5.0.tar.gz' is not qualified and no path was specified. Please qualify the command or specify a path.

URLL is good, curl is installed. I don't see why puppet can't download the archive.

I will look deeper into sources latter inthe week to see if I can find some more informations.

This provider doesn't work on OSX since it requires Package['curl'] as a dependency, even so
the 'curl' program is distributed as part of the base OS install and no special package exists.

Hi
Didn't found any information about that. So my question are you planing to support puppet:/// as url?

We have an archive resource like this:

   archive { 'oracle_patch':
      ensure        => present,
      url           => 'http://some.server/path/oracle_patch.tgz',
      target        => '/some/path/local',
      digest_type   => 'md5',
      digest_string => $::oracle_patch_md5sum,
      extension     => 'tgz',
      src_target    => '/some/path/local',
      checksum      => true,
      user          => 'oracle',
      timeout       => 600,
    }

$::oracle_patch_md5sum is defined in hiera.

First run everything is fine, archive is downloaded and extracted.
So days later we updated the archive on the webserver and the md5sum in hiera.

The md5sum is created correctly as the new one, but the file isn't downloaded again, because the exec-resource is only firing if no downloaded file is present (see 'creates' below).

      exec {"download archive ${name} and check sum":
        command     => "curl ${proxy_option} -s -S ${insecure_arg} ${redirect_arg} -o ${src_target}/${name} '${url}'",
        creates     => "${src_target}/${name}",
        logoutput   => true,
        timeout     => $timeout,
        path        => $path,
        require     => Package['curl'],
        notify      => $_notify,
        user        => $user,
        refreshonly => $refreshonly,
      }

So the checksum checks went wrong and everything is deleted.
On the second puppet run everything is running fine, since nothing is present on local disk.

Modify mkdir command execution to support execution user and owner.

exec {"$name unpack":
        command => $extension ? {
          'zip'     => "mkdir -p ${target} && ${extract_zip}",
          'tar.gz'  => "mkdir -p ${target} && ${extract_targz}",
          'tgz'     => "mkdir -p ${target} && ${extract_targz}",
          'tar.bz2' => "mkdir -p ${target} && ${extract_tarbz2}",
          'tgz2'    => "mkdir -p ${target} && ${extract_tarbz2}",
          default   => fail ( "Unknown extension value '${extension}'" ),
        },
        creates => $extract_dir,
        timeout => $timeout
      }

Currently all archives are owned by root

There doesn't currently appear to be a means of specifying basic auth credentials for downloads; is this something that would be a good addition to the module?

Trying to download a file from a URL that doesn't exist fails silently and running puppet agent -t appears as if everything works correctly. According to e the cURL man page, if you add a -f to the exec command in download.pp, it will return error code 22 upon most 4xx error codes. Are there implications that I'm not seeing that would prevent something like this from being implemented?

I ran into an issue where the source file was downloaded incorrectly. I specified the wrong URL but curl was able to download the server's response. Subsequent puppet agent runs would not re-download the file.

Is there an option to re-download a file? or delete a file after it has been downloaded?

if we "install" a new version of an archive, an unpredictable amount of files may have changed.
in such a case, it may be necessary to purge the target directory prior to extraction.

Getting the error: 'curl is not qualified and no path was specified. Please qualify the command or specify a path. I would love to specify the path though. Maybe a misconfiguration of my puppet env?

After upgrading from 0.5.x to 0.6.1 we have
Archive::Extract[liquibase-3.3.2]/Exec[liquibase-3.3.2 unpack]/returns (err): change from notrun to 0 failed: Could not find command ''

I need to set particular permissions on an extracted directory. But on every puppet run, archive extract it once again and overide permissions that I try to maintain...

archive { 'gaussian':
    ensure    =>  present,
    url       =>  'http://myfileserver/sources/gaussian_09.tgz',
    target    =>  '/opt',
    extension => "tgz",
    checksum  =>  false,
  }

file { '/opt/gaussian':
    ensure  =>  directory,
    owner   =>  "root",
    group   =>  "users",
    recurse =>  true,
    mode    =>  "0750",
    require => Archive['gaussian'],
}

Replace direct usage of ${name}. Use some other field to pass the name of archive. Currently scenario where there is need to extract one archive to multiple destination is not possible. Maybe something like:

define archive::extract (
  $archive_name = $title,
  $target,
  $ensure=present,
  $src_target='/usr/src',
  $root_dir='',
  $extension='tar.gz',
  $timeout=120) { ...

It should run once, when files do not exist in $target_dir. however, it always runs. Why is this?