[document] fix and improve the current docs

[mocchira]

Fixes suggested at https://groups.google.com/forum/?utm_medium=email&utm_source=footer#!msg/leoproject_leofs/QsxY2wNWhYg/YV2Xlx2AFgAJ

• [ ] Fix wrong explanations
  • https://leo-project.net/leofs/docs/getting_started/getting_started_2.html to be fixed as manager.partner is required
  • installation guide to be fixed to recommend the package installation rather than the source one
  • the consistency level(R:W:D) except num_of_replicas(N) can be changed on the fly
    • the procedure for temporal change would be useful
    • the procedure for parmanent change would be useful
  • http://leo-project.net/leofs/docs/_images/leofs-upgrade-flow-diagram.jpg
    • currently diagram shows "copy, then stop, then start" but should be "stop then copy, then start"
• [ ] Improvements (more strict explanation)
  • consistency.num_of_replicas
    • what happend if changed?
  • consistency.rack_aware_replicas
  • ports opened on managers
  • obj_containers.(paths|num_of_containers)
    • what users have to do if they want to change those?
  • metadata (how much disk space do those take? and so on)
    • mnesia
    • queue
    • metadata corresponding to each object
  • http://leo-project.net/leofs/docs/faq/faq_3.html#how-do-i-set-a-number-of-containers-at-leofs-storage-configuration list advantages/disadvantages as much as possible
    • small number
    • big one
  • Erlang VM related properties
    • when user have to change those?
  • num_of_vnodes
    • how to determine the appropriate value?
    • when user have to change?
      • solve deviation of disk space per node
  • Pitfalls when running multiple gateways in different mode
  • Administration Guide: LeoFS Storage Operations.
    • Compaction on old nodes might be highly recommended after rebalance is finished.
      • after you add a few storage nodes, disk usage on existing nodes won't go down until compaction is performed.