Comment 4 for bug 1333584

Reviewed: https://review.openstack.org/104567
Committed: https://git.openstack.org/cgit/stackforge/fuel-web/commit/?id=d01b4efc0fc4af9d0e316b9dfc7974f16975f822
Submitter: Jenkins
Branch: stable/5.0

commit d01b4efc0fc4af9d0e316b9dfc7974f16975f822
Author: Evgeniy L <email address hidden>
Date: Wed Jul 2 21:09:53 2014 +0400

    Upgrades, changed workflow for versions.yaml config file

    Now we have the next worflow
    * if user runs upgrade script first time,
      we save current version.yaml file into
      work directory
    * if user for some reaons runs the script
      second or third time, we use version.yaml
      from work directory as the version which
      we need to rollback to, we cannot use
      config from /etc/fuel/version.yaml.
      Because upgrade script could be
      interrupted after symlink creation

    Additional fixes
    * config has version parameters which looks
      like strings instead of dicts with hierarchy,
      because we don't need all of the data from
      version.yaml file
    * removed before and post upgrade section
      because they were useless, in fact we
      can do this actions in upgrade method
    * created before upgrade checker which is
      activated only if the script was ran
      with docker parameter, this checker
      checks if ugprade make sense. For example
      now impossible to run upgrade from version
      5.0 to 5.0 and from 5.1 to 5.0.1

    blueprint upgrade-to-5-0-1
    Closes-bug: #1331363
    Closes-bug: #1333584
    Change-Id: I719799db216260d527ce5e9be2f03acc234cf3b0