Puppet Run giving Error: failed to execute puppetrun: ERF12-4252

See what the command actually returns when run as shell from root by running echo $? after running the actual command, probably it returns not 0. If so you need to change the statement to:

exec {"initiate replica set":
    command => "echo 'rs.initiate()' | mongo",
    path => ["/usr/bin","/usr/sbin","/bin"],
    user => "root",
    require => Class["mongodb"],
    returns => $ActualReturnCode
}

Where $ActualReturnCode is the code that you get from echo $? after running the command as root.

There is one more thing about mongodb I forgot to mention - when you install the package it will take some time for the main mongodb database to init, until then you won't be able to access it. Thus the workaround here would be to also add sleep and a number of retries making the code look like:

exec {"initiate replica set":
    command => "echo 'rs.initiate()' | mongo",
    path => ["/usr/bin","/usr/sbin","/bin"],
    user => "root",
    require => Class["mongodb"],
    retries => 5,
    sleep   => 60,
}

Final Edit

Not sure what I did to get it to work last time but the true cause of my misfortune was omitting this step,

http://docs.openstack.org/havana/install-guide/install/apt/content/basics-packages.html

which guides you through the process of installing the havana ubuntu cloud archive.

Thanks for reading!

old post starts here

I managed to fix the issue.

In the keystone database, the endpoints table more specifically there was a mix of localhoist/and ip address based entries in the "url" field, they look like this x.x.x.x:9292, x.x.x.x:8776/v1/%(tenant_id)s and so on. I replaced the ip address with localhost in all cases making it uniform and I'm now able to log into the dashboard from the machine itself. I can't login other than from localhost but I'm assuming this is a separate issue because I tried replacing localhost with other entries (the public ip address for example and I can still login via dashboard from localhost).

To summarize, although I could be misinterpreting what happened, I think the fact that the endpoints url fields weren't uniform was causing the initial connection refused issue. I believe the specific endpoints that were written out as localhost caused the problem, unfortunately I didn't take note of which were which at the time.

Edit: I tried accessing the dashboard from my browser in a incognito window using chrome and I could log in no problem, I tried internet explorer and it works great. But even after clearing out all my history/cache/etc from chrome I still get the error. Not sure why but at least I've narrowed things down a bit, and it works, kinda!