Getting Started with Ceilometer and Gnocchi


Introduction

With our latest update to Private Cloud Core comes two new OpenStack services: Ceilometer and Gnocchi. These services are part of OpenStack’s Telemetry Project and serve to collect data. That data then can be acted upon using the OpenStack alarming service Aodh.

The purpose of this guide is to briefly explain the function of Ceilometer, Gnocchi, and Aodh. We also demonstrate a way to collect instance resource usage metrics provided by Ceilometer using OpenStackClient.


 

What is Ceilometer?

Ceilometer is an OpenStack service that provides cloud metrics. The data this service provides can be used for customer billing, resource usage analysis,and to send alerts.

Reference: Ceilometer Documentation

What is Gnocchi?

Gnocchi, part of the bigger Telemetry project in OpenStack, addresses the problem of storage and collection of time series data. Gnocchi provides time series database as a service.

Reference: Gnocchi Wiki

What is Aodh?

Aodh is an OpenStack alarming service that can provide alerts and trigger actions based on the metrics collected by Ceilometer and Gnocchi.

Reference: Aodh Documentation

 

How can I work with Ceilometer and collect metrics?

This section describes software that makes use of Ceilometer and Gnocchi and also a way to collect instance usage metric data.


An example of software that can be installed to an OpenStack cloud is Fleio. This is software that makes use of metric collecting services like Ceilometer and Gnocchi to provide a billing system and self-service portal. 

Note! — Fleio is third party software and is not part of the default software installed to a Private Cloud Core cloud.

 

Collect Instance Metrics using OpenStackClient

Using the Ceilometer command line client, you can collect instance resource usage metrics like CPU time, memory, and disk space used.

This section demonstrates how you can use OpenStackClient to collect instance resource usage metrics.

Required Command Line Clients

Before proceeding, the required command line clients must be installed.

To work with Gnocchi, Ceilometer, and Aodh over the command line, the following packages must be installed to your OpenStackClient virtual environment:

  • gnocchiclient
  • python-ceilometerclient
  • aodhclient

 

How to Collect Instance Metrics from Ceilometer

Step 1 — Install required command line client

This section requires python-ceilometerclient to be installed and an OpenStack cloud that has at least one instance using resources within thecloud.

To install this package, ensure you have a prepared OpenStackClient environment, then install the package using pip.

For example:

pip install python-ceilometerclient

 

Step 2 — List instances

Use openstack server list to list instances.

For example:

$ openstack server list
+--------------------------------------+----------------------------+--------+---------------------------------------+------------------------------------------------------+----------+
| ID                                   | Name                       | Status | Networks                              | Image                                                | Flavor   |
+--------------------------------------+----------------------------+--------+---------------------------------------+------------------------------------------------------+----------+
| 1d6e3cd6-2a2d-4603-b854-47966e4d0eef | instance-1                 | ACTIVE | demo-net=173.231.254.73, 192.168.1.87 | N/A (booted from volume)                             | c1.small |
| 52e3a93f-a593-404e-91fd-53ee7ebc6d86 | test-4j5gttlkz52p-node-2   | ACTIVE | test=10.0.0.233, 173.231.254.75       | Fedora CoreOS (fedora-coreos-33.20210412.3.0-stable) | m1.small |
| e4d67716-4d1a-4460-9276-bd0da12f3cca | test-4j5gttlkz52p-node-1   | ACTIVE | test=10.0.0.143, 173.231.254.70       | Fedora CoreOS (fedora-coreos-33.20210412.3.0-stable) | m1.small |
| ef263252-54a7-40a9-9d6a-14814c5ed66c | test-4j5gttlkz52p-node-0   | ACTIVE | test=10.0.0.188, 173.231.254.78       | Fedora CoreOS (fedora-coreos-33.20210412.3.0-stable) | m1.small |
| ccea9641-879e-434d-ab69-f538dc291d34 | test-4j5gttlkz52p-master-2 | ACTIVE | test=10.0.0.148, 173.231.254.71       | Fedora CoreOS (fedora-coreos-33.20210412.3.0-stable) | m1.small |
| 3f108ea3-12bc-494c-83bf-16f3be3f839a | test-4j5gttlkz52p-master-1 | ACTIVE | test=10.0.0.88, 173.231.254.76        | Fedora CoreOS (fedora-coreos-33.20210412.3.0-stable) | m1.small |
| b989247d-a879-4ad9-bdb7-dae219039b33 | test-4j5gttlkz52p-master-0 | ACTIVE | test=10.0.0.117, 173.231.254.72       | Fedora CoreOS (fedora-coreos-33.20210412.3.0-stable) | m1.small |
+--------------------------------------+----------------------------+--------+---------------------------------------+------------------------------------------------------+----------+

From this output, you need the instance’s UUID. Copy this for the next step.

 

Step 3 — List metrics for an instance

With the instance UUID, use openstack metric list -c resource_id -c name -c
unit | grep <instance-uuid>
to list metrics.

For example:

$ openstack metric list -c resource_id -c name -c unit | grep 1d6e3cd6-2a2d-4603-b854-47966e4d0eef
| cpu                             | ns      | 1d6e3cd6-2a2d-4603-b854-47966e4d0eef |
| memory.usage                    | MB      | 1d6e3cd6-2a2d-4603-b854-47966e4d0eef |
| vcpus                           | vcpu    | 1d6e3cd6-2a2d-4603-b854-47966e4d0eef |
| compute.instance.booting.time   | sec     | 1d6e3cd6-2a2d-4603-b854-47966e4d0eef |
| disk.ephemeral.size             | GB      | 1d6e3cd6-2a2d-4603-b854-47966e4d0eef |
| memory                          | MB      | 1d6e3cd6-2a2d-4603-b854-47966e4d0eef |
| disk.root.size                  | GB      | 1d6e3cd6-2a2d-4603-b854-47966e4d0eef |

Listed in the left-most column are the names of the metrics. The column immediately following this one contains the units in which the metrics are measured.

 

Step 4 — List CPU usage metrics

This step demonstrates collecting the CPU usage for an instance.

To collect CPU usage, use openstack metric measures show --resource-id=<instance-uuid> cpu.

For example:

$ openstack metric measures show --resource-id=1d6e3cd6-2a2d-4603-b854-47966e4d0eef cpu
+---------------------------+-------------+----------------+
| timestamp                 | granularity |          value |
+---------------------------+-------------+----------------+
| 2021-09-09T18:00:00-04:00 |       300.0 |  20340000000.0 |
| 2021-09-09T18:05:00-04:00 |       300.0 |  20680000000.0 |
| 2021-09-09T18:10:00-04:00 |       300.0 |  21010000000.0 |
| 2021-09-09T18:15:00-04:00 |       300.0 |  21440000000.0 |
| 2021-09-09T18:20:00-04:00 |       300.0 |  21880000000.0 |
| 2021-09-09T18:25:00-04:00 |       300.0 |  22290000000.0 |
| 2021-09-09T18:30:00-04:00 |       300.0 |  22590000000.0 |
| 2021-09-09T18:35:00-04:00 |       300.0 |  22870000000.0 |
[... output truncated...]

 

Conclusion

Beyond this, Aodh can be used to act on metrics, by providing alarms or causing specific actions to occur. For more on setting up alarms using Aodh, see Using Alarms, from the OpenStack documentation page. Setting up alarms or causing specific actions to occur is beyond the scope of this guide.

NW
Nick West Systems Engineer

Nick is an avid aggressive inline skater, nature enthusiast, and loves working with open source software in a Linux environment.

More Articles by Nick

Was this article helpful? Let us know!