netring fact is broken because no waiting cycle for running threads

[Sergey Vasilenko]

Detailed bug description:
  test spec/unit/facter/netrings_spec.rb has false-negative result sometime.

Steps to reproduce:
  I found two independed methods for repro:
  1. Run following scenario under OSX multiple times
     # cd deployment/puppet/l23network
     # rake spec
  2. create huge of pseudo-interfaces, subinterfaces or linux bridges,
     run following scenario:
     # cd deployment/puppet/l23network
     # rake spec
     breal running tests and run test
     # rspec spec/unit/facter/netrings_spec.rb
     multiple times

Expected results:
 all tests should be green

Actual result:
 test spec/unit/facter/netrings_spec.rb failed

Reproducibility:
 100% on OSX
 ~40% on linux

Impact:
 l23network puppet module

[Sergey Vasilenko (xenolog)]

While I research this bug, I found, that no waiting cycle for result of all running threads.

[Sergey Vasilenko (xenolog)]

I mark this bug as High for Mitaka, because threading issue is important

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/323386

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/323386
Committed: https://git.openstack.org/cgit/openstack/fuel-library/commit/?id=bb134b2e04c1c385cf8ce4d0c753808750ca9122
Submitter: Jenkins
Branch: master

commit bb134b2e04c1c385cf8ce4d0c753808750ca9122
Author: Sergey Vasilenko <email address hidden>
Date: Tue May 31 16:39:47 2016 +0300

    Refactor Netrings fact and it's tests for:

    1. passing test if run under OSX (stub :kernel fact into test)
    2. make waiting cycle for all started threads
    3. re-implement netrings properties gather as ruby class

    Change-Id: I4398804040c819b783cebd05842ffdbbb8ea561e
    Closes-bug: #1587470

[Sergey Vasilenko (xenolog)]

I mark this bug "Won't fix" for Mitaka series, because code from https://bugs.launchpad.net/fuel/+bug/1456723 was not merged into stable/mitaka branch.