rabbitmq cluster goes ACTIVE but is unusable after initial boot

[Min Pae]

The rabbitmq cluster, when booted, starts up rabbitmq services to the point where the cluster goes ACTIVE, but the cluster becomes unusable after some time (~3 minutes in the cluster where this was observed, but not quantified beyond that). It seems like the service is being reset and restarted several times beyond the initial time it is "supposed" to be reset/restarted. From observation the os-collect-config service is still running even though it is being disabled in the post-configure step. This would explain the repeated reset/restarts observed in the rabbitmq logs.

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/181610
Committed: https://git.openstack.org/cgit/stackforge/cue/commit/?id=f91bc8965b50c212778f857006899f2b90325e0a
Submitter: Jenkins
Branch: master

commit f91bc8965b50c212778f857006899f2b90325e0a
Author: Min Pae <email address hidden>
Date: Fri May 8 21:42:57 2015 -0700

    configure and start rabbitmq VMs from userdata

    Configuring and starting rabbitmq with os-apply-config and
    os-refresh-config seems to be causing issues where rabbitmq is being
    regularly reset and restarted. We want rabbitmq to be reset and started
    once at initial boot and not past that. Moving the config generation
    and service startup logic to the userdata script that is supplied by Cue
    at VM creation time.

    Closes-Bug: 1453351

    Change-Id: I2971c196e0cc106d17e18f868c491b589ecb17b8