Puppet provision fails on yum upgrade

[shentonfreude]

[vagrant@vagrant-centos-64 yum.repos.d]$ sudo puppet apply -v /tmp/vagrant-puppet/manifests/sid_ldap.pp 

err: Could not prefetch package provider 'yum': Execution of '/usr/bin/python /usr/lib64/ruby/gems/1.8/gems/puppet-2.6.4/lib/puppet/provider/package/yumhelper.py' returned 1: Loaded plugins: fastestmirror
Loading mirror speeds from cached hostfile
 * base: centos.mirror.nac.net
 * extras: mirrors.greenmountainaccess.net
 * rpmforge: fr2.rpmfind.net
 * updates: mirror.trouble-free.net
Traceback (most recent call last):
  File "/usr/lib64/ruby/gems/1.8/gems/puppet-2.6.4/lib/puppet/provider/package/yumhelper.py", line 115, in ?
    ypl = pkg_lists(my)
...

  File "/usr/lib/python2.4/site-packages/yum/yumRepo.py", line 1395, in _getRepoXML
    raise Errors.RepoError, msg
yum.Errors.RepoError: Cannot retrieve repository metadata (repomd.xml) for repository: elff. Please verify its path and try again

[shentonfreude]

It looks like I can't even fix this by hiding the bogus repo in question (or fiddling the hostname in /etc/hosts) because it seems Puppet is running a "python ... .../yumhelper.py" before starting on any of my defined Stages:

...
[default] debug: Puppet::Type::Package::ProviderYum: Executing '/bin/rpm -qa --nosignature --nodigest --qf '%{NAME} %|EPOCH?{%{EPOCH}}:{0}| %{VERSION} %{RELEASE} %{ARCH}
''
[default] 
[default] debug: Puppet::Type::Package::ProviderYum: Executing '/usr/bin/python /usr/lib64/ruby/gems/1.8/gems/puppet-2.6.4/lib/puppet/provider/package/yumhelper.py'
[default] 
[default] err: Could not prefetch package provider 'yum': Execution of '/usr/bin/python /usr/lib64/ruby/gems/1.8/gems/puppet-2.6.4/lib/puppet/provider/package/yumhelper.py' returned 1: Loaded plugins: fastestmirror
Determining fastest mirrors
...

yum.Errors.RepoError: Cannot retrieve repository metadata (repomd.xml) for repository: elff. Please verify its path and try again

[default] 
[default] debug: /Stage[main]/Trac_instance/File[/var/trac/project1/db]/require: requires Exec[trac_create_project1]
...

[shentonfreude]

Only solution I can see is:

1. Hack for now: just run it twice and get on with the real work
2. find or BUILD a Vagrant image of Centos-64
   • help to build your own: https://gist.github.com/1052439

[shentonfreude]

I looked at VeeWee but the boxes it knew about didn't have a 64-bit Centos 5.6 or 5.7. Since SID is using RHEL 5.7, we need something similar for credible development.

Looks like it's time to build our own.

This is even more reasonable since vagrantbox.es hasn't been reachable the past hour.

There's a commandline interface to search vagrantbox.es:

gem install vagrantboxes

But I couldn't get the subsequent command to work like it wasn't registered with vagrant properly:

vagrant vagrantboxes search

I'm not going down that rathole now.