ceilometer-central-agent and ceilometer-alarm-evaluator services have stopped status in pcs on HA mode

[Vadim Rovachev]

{"build_id": "2015-01-31_22-55-01", "ostf_sha": "c9100263140008abfcc2704732e98fbdfd644068", "build_number": "90", "release_versions": {"2014.2-6.1": {"VERSION": {"build_id": "2015-01-31_22-55-01", "ostf_sha": "c9100263140008abfcc2704732e98fbdfd644068", "build_number": "90", "api": "1.0", "nailgun_sha": "c0932eb5c2aa7fd1e13a999cb1b4bf5aff101c3b", "production": "docker", "python-fuelclient_sha": "2ea7b3e91c1d2ff85110bf5abb161a6f4e537358", "astute_sha": "ed5270bf9c6c1234797e00bd7d4dd3213253a413", "feature_groups": ["mirantis"], "release": "6.1", "fuelmain_sha": "", "fuellib_sha": "c5e4a0410ba66f9e9911f62b3b71c0b9c29aed6e"}}}, "auth_required": true, "api": "1.0", "nailgun_sha": "c0932eb5c2aa7fd1e13a999cb1b4bf5aff101c3b", "production": "docker", "python-fuelclient_sha": "2ea7b3e91c1d2ff85110bf5abb161a6f4e537358", "astute_sha": "ed5270bf9c6c1234797e00bd7d4dd3213253a413", "feature_groups": ["mirantis"], "release": "6.1", "fuelmain_sha": "", "fuellib_sha": "c5e4a0410ba66f9e9911f62b3b71c0b9c29aed6e"}

[root@node-1 ~]# pcs resource | grep ceilo
 p_openstack-ceilometer-central (ocf::fuel:ceilometer-agent-central): Stopped
 p_openstack-ceilometer-alarm-evaluator (ocf::fuel:ceilometer-alarm-evaluator): Stopped

ceilometer-central-agent and ceilometer-alarm-evaluator services are configured under pacemaker but not monitored and run as upstart jobs. We don't need in replica for these services because it's enough to have one central(without workload partitioning as we are having now) and alarm-evaluator agents per cluster.

So i think we should enable the monitoring by pacemaker.

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/154361

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: stable/6.0
Review: https://review.openstack.org/154906

[Daniil Trishkin (dtrishkin)]

It does not make sense to have started ceilometer-central-agent without coordination, also it does not make sense to have started ceilometer-alarm-evaluator more than one per claster. This is disoriented users only. So here we enabled this services via pacemaker. This change should to be present in 6.0 too.

[Daniil Trishkin (dtrishkin)]

1. Create env with HA mode
       Nodes: (controller + mongo) * 3
2. Wait for end of deploy
3. Type: pcs resource

You'll see different resources started via pacemaker, also here will be ceilometer-central-agent and ceilometer-alarm-evaluator with "Started" status.

Failover: If you force shutdown node with this agents, they'll migrate to other node in cluster. After that you can check this by the same command: pcs resource, and you'll see "Started" status opposite this two resources.

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/154361
Committed: https://git.openstack.org/cgit/stackforge/fuel-library/commit/?id=8daf024982265a19fbc552e1181685e5dd1a641d
Submitter: Jenkins
Branch: master

commit 8daf024982265a19fbc552e1181685e5dd1a641d
Author: dtrishkin <email address hidden>
Date: Tue Feb 10 10:12:35 2015 +0300

    Start ceilometer resource agents

     * Ceilometer agent central and ceilometer alarm evaluator
       now start as resource agents of pacemaker;
     * Disable upstart for this two services.

    Change-Id: Ic15a3bfb6238e4281b06aae0a3f9fe4abf96590d
    Closes-bug: #1418984
    Closes-bug: #1422403

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/154906
Committed: https://git.openstack.org/cgit/stackforge/fuel-library/commit/?id=ba1b2e5f2e763df0585a94f6cf04d582aa545221
Submitter: Jenkins
Branch: stable/6.0

commit ba1b2e5f2e763df0585a94f6cf04d582aa545221
Author: dtrishkin <email address hidden>
Date: Tue Feb 10 10:12:35 2015 +0300

    Start ceilometer resource agents

     * Ceilometer agent central and ceilometer alarm evaluator
       now start as resource agents of pacemaker
     * Disable upstart for this two services

     It does not make sense to have started ceilometer-central-agent
     without coordination, also it does not make sense to have
     started ceilometer-alarm-evaluator more than one per claster.
     This is disoriented users only. So here we enabled
     this services via pacemaker. This change should be
     present in 6.0 too.

    DocImpact

    Change-Id: Ic15a3bfb6238e4281b06aae0a3f9fe4abf96590d
    Closes-bug: #1418984
    Closes-bug: #1422403

[Anastasia Palkina (apalkina)]

Verified on ISO #202

"build_id": "2015-03-16_22-54-44", "ostf_sha": "e86c961ceacfa5a8398b6cbda7b70a5f06afb476", "build_number": "202", "release_versions": {"2014.2-6.1": {"VERSION": {"build_id": "2015-03-16_22-54-44", "ostf_sha": "e86c961ceacfa5a8398b6cbda7b70a5f06afb476", "build_number": "202", "api": "1.0", "nailgun_sha": "874df0d06e32f14db77746cfeb2dd74d4a6e528c", "production": "docker", "python-fuelclient_sha": "2509c9b72cdcdbe46c141685a99b03cd934803be", "astute_sha": "93e427ac49109fa3fd8b0e1d0bb3d14092be2e8c", "feature_groups": ["mirantis"], "release": "6.1", "fuelmain_sha": "608b72a6f79a719cf01c35a19d0091fe20c8288a", "fuellib_sha": "924d73ae4766646e1c3a44d7b59c4120985e45f0"}}}, "auth_required": true, "api": "1.0", "nailgun_sha": "874df0d06e32f14db77746cfeb2dd74d4a6e528c", "production": "docker", "python-fuelclient_sha": "2509c9b72cdcdbe46c141685a99b03cd934803be", "astute_sha": "93e427ac49109fa3fd8b0e1d0bb3d14092be2e8c", "feature_groups": ["mirantis"], "release": "6.1", "fuelmain_sha": "608b72a6f79a719cf01c35a19d0091fe20c8288a", "fuellib_sha": "924d73ae4766646e1c3a44d7b59c4120985e45f0"