HI Pierre,

You are correct it was my mistake to run playbook without -i multinode so it just ran on the controller node. Now I can see the ceilometer has been deployed on the compute nodes. But i can't see data in cloudkitty so I'm trying to debug. 

My understanding is that a ceilometer pushes the data to gnocchi and cloudkitty consumes it using a rating engine. I am able to see data in gnocchi but cloudkitty showing all zero cost in GUI. I did configure rates based on flavor_id etc from openstack example. Not seeing any interesting error or log data which indicate problem anywhere. 

# openstack metric measures show e7913002-5f7b-4229-8cd8-61577a4fcf73
+---------------------------+-------------+--------------+
| timestamp                 | granularity |        value |
+---------------------------+-------------+--------------+
| 2022-12-23T03:55:00+01:00 |       300.0 | 891.00390625 |
| 2022-12-23T04:00:00+01:00 |       300.0 | 890.97265625 |
| 2022-12-23T04:05:00+01:00 |       300.0 | 891.00390625 |
| 2022-12-23T04:10:00+01:00 |       300.0 |  890.7890625 |
| 2022-12-23T04:15:00+01:00 |       300.0 | 891.00390625 |

On Sat, Dec 24, 2022 at 9:45 AM Pierre Riteau <pierre@stackhpc.com> wrote:
With these settings enabled, you should see a ceilometer_compute container on each hypervisor. Try deploying/reconfiguring without any limit?

On Thu, 22 Dec 2022 at 23:21, Satish Patel <satish.txt@gmail.com> wrote:
Folks,

I am deployed kolla-ansible and trying to set up telemetry for billing etc. But not able to get it working may be broken or something. 

In global.yml I turned on the following options and deployed kolla and it did install all the components but I am not seeing it deploy a ceilometer agent on compute nodes. How does it collect data from compute nodes if the ceilometer agent is not running on compute nodes? Did I miss something here? 

enable_ceilometer: "yes"
enable_gnocchi: "yes"
enable_cloudkitty: "yes"