LSB facts are a pain.

[razorsedge]

@jhoblitt

When installing @core on EL6 during kickstart, redhat-lsb is not installed resulting in the following on the first Puppet run:

Module ipmi is not supported on lsbmajdistrelease at /etc/puppetlabs/puppet/modules/forge/ipmi/manifests/params.pp:26

I try not to use LSB facts, and will throw the following in the params.pp and reference $::module::params::majdistrelease in the module code:

  if $::operatingsystemmajrelease { # facter 1.7+
    $majdistrelease = $::operatingsystemmajrelease
  } elsif $::lsbmajdistrelease {    # requires LSB to already be installed
    $majdistrelease = $::lsbmajdistrelease
  } elsif $::os_maj_version {       # requires stahnma/epel
    $majdistrelease = $::os_maj_version
  } else {
    $majdistrelease = regsubst($::operatingsystemrelease,'^(\d+)\.(\d+)','\1')
  }

[jhoblitt]

The facter RPM probably should pull in redhat-lsb as a dep.

I'm not enthusiastic about putting that logic into the module as it's working around what is really a fact problem. I'd rather do something line checking for the defined-ness of the used lsb fact and failing if it's not set, introducing a dep on a module that will install the redhat-lsb package, or using ensure_packages() from stdlib to make sure redhat-lsb is present.

[razorsedge]

I firmly believe that facter should not require redhat-lsb. That's a whole load of software many people do not want on their servers (including CUPS).

The problem here is that I actually do include lsb in my manifests along with include ipmi, but catalog compilation fails. Perhaps you could use $::operatingsystemmajrelease instead of $::lsbmajdistrelease?

Or do you really need to fail on the default of case $::lsbmajdistrelease { }?