[Ceph] Do not restart Ceph service on new Ceph OSD node deployment

[Miroslav Anashkin]

The following improvement greatly increased the OSD deployment robustness
https://github.com/stackforge/fuel-library/commit/c20b2f8463e962734222382a77f26b91a53fd264

However, there is still one issue:
https://github.com/stackforge/fuel-library/blob/c20b2f8463e962734222382a77f26b91a53fd264/deployment/puppet/ceph/manifests/init.pp#L136

`Class['ceph::conf'] -> Class['ceph::osds'] ~> Service['ceph']`

`Class ceph::osds::osd` does both for each OSD , `ceph deploy osd prepare` and `ceph deploy osd activate`.

After that, Puppet code sends notification to Service['ceph']
As a result, Service['ceph'] is being restarted, followed by all deployed OSD restart on the entire node.
This step is not necessary, since all the related Ceph OSD services has been already started by `ceph deploy osd activate`.
It only creates huge load (especially in cases when OSD is added to long-existing cluster under the load) and may lead to unstable errors when particular OSD services fail to start.

In order to remove this service restart please change the problematic code line as following:
Class['ceph::conf'] -> Class['ceph::osds'] -> Service['ceph']

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/207761

[Vitaly Sedelnik (vsedelnik)]

We don't support master node updates in 5.1/5.1.1 and 6.0 maintenance updates. Setting to Invalid for 5.1.1 and 6.0 updates.

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/207761
Committed: https://git.openstack.org/cgit/stackforge/fuel-library/commit/?id=141f6f7ed7039644b5e451cc521f2fc492ae6e81
Submitter: Jenkins
Branch: master

commit 141f6f7ed7039644b5e451cc521f2fc492ae6e81
Author: Bogdan Dobrelya <email address hidden>
Date: Fri Jul 31 09:47:37 2015 +0200

    Restart ceph service only on conf change

    W/o this patch, puppet sends notification
    to Service['ceph'] unconditionally. As a
    result, Service['ceph'] is being restarted,
    followed by all deployed OSD restart on the
    entire node.

    This is a problem since all the related
    Ceph OSD services has been already started
    by `ceph deploy osd activate` and this notify
    with restart only creates huge load
    (especially in cases when OSD is added to
    long-existing cluster under the load) and may
    lead to unstable errors when particular OSD
    services fail to start.

    The solution is to trigger ceph service restart
    only on configuration updates.

    Closes-bug: #1479880

    Change-Id: Ib60436d50379041a6a1368b0fa170b9b58cc484c
    Signed-off-by: Bogdan Dobrelya <email address hidden>

[Oleksiy Molchanov (omolchanov)]

Verified

VERSION:
  feature_groups:
    - mirantis
  production: "docker"
  release: "7.0"
  openstack_version: "2015.1.0-7.0"
  api: "1.0"
  build_number: "284"
  build_id: "284"
  nailgun_sha: "5c33995a2e6d9b1b8cdddfa2630689da5084506f"
  python-fuelclient_sha: "1ce8ecd8beb640f2f62f73435f4e18d1469979ac"
  fuel-agent_sha: "082a47bf014002e515001be05f99040437281a2d"
  fuel-nailgun-agent_sha: "d7027952870a35db8dc52f185bb1158cdd3d1ebd"
  astute_sha: "8283dc2932c24caab852ae9de15f94605cc350c6"
  fuel-library_sha: "f81fdabe6c05be7a3d11d88a7c3a8f3931921c73"
  fuel-ostf_sha: "1f08e6e71021179b9881a824d9c999957fcc7045"
  fuelmain_sha: "9ab01caf960013dc882825dc9b0e11ccf0b81cb0"