[Cinder] service_down_time value less then report_interval

[Leontii Istomin]

Detailed bug description:
"cinder-manage service list" shows that cinder services are flopping
http://paste.openstack.org/show/495645/
Actually all services are running and I can create a volume without issues
I've found that in [DEFAULT] section of /etc/cinder/cinder.conf:
report_interval = 60
service_down_time = 10
It means that cinder services report about their status too rarely.
I believe we need to set report_interval parameter to 5 if we would like to check the services once per 10 seconds
Steps to reproduce:
1. Deploy Fuel 9.0-244
2. Deploy cluster: 3 controllers, 20 computes+Ceph, 175 computes, Ceph as a backend for all
3. Perform "cinder-manage service list"
Expected results:
The command shows all services with ":-)"
Actual result:
The command shows all services with "XXX"
Reproducibility:
Each time
Workaround:
report_interval parameter to 5 (Didn't check)
Impact:
Monitoring services like LMA
Description of the environment:
- Operation system: Ubuntu
- Versions of components: MOS9.0
- Reference architecture: 3 controllers, 20 computes+Ceph, 175 computes, Ceph as a backend for all
- Network model: vxlan+DVR
- Related projects installed:
Additional information:

[Swann Croiset (swann-w)]

fyi, related bug https://bugs.launchpad.net/fuel/+bug/1448151

[Roman Podoliaka (rpodolyaka)]

MOS Puppet, could you please check the Puppet manifests settings?

[Alexey Deryugin (velovec)]

Fix already merged: https://review.openstack.org/#/c/310469/