7.8. Enabling Metering for Compute Resources

You can collect usage data of compute resources using Gnocchi. It is a time series database that processes and stores measurement data of compute resources and provides access to it via REST API or the command-line tool.

Gnocchi aggregates and stores measures for compute resource metrics according to their archive policy. The policy defines how long aggregates are kept in a metric and how they are computed (minimum, maximum, average, etc.). The archive policy is set when a metric is created and cannot be changed afterwards. All the existing metrics have the low archive policy, meaning that all computed aggregates are kept for one day with 5-minute granularity and one month with 1-hour granularity.

The following metrics are available for aggregation:

Table 7.8.1 Compute resource metrics
Resource Metric Type Description
Virtual machine memory gauge Amount of RAM allocated to the VM, in megabytes
vcpus gauge Number of virtual CPUs allocated to the VM
disk.read.requests cumulative Number of read requests
disk.write.requests cumulative Number of write requests
disk.read.bytes cumulative Amount of data read, in bytes
disk.write.bytes cumulative Amount of data written, in bytes
network.incoming.bytes cumulative Incoming network traffic, in bytes
network.outgoing.bytes cumulative Outgoing network traffic, in bytes
network.incoming.packets cumulative Incoming network traffic, in packets
network.outgoing.packets cumulative Outgoing network traffic, in packets
Image image.size gauge Size of the uploaded image, in bytes
Volume volume.size gauge Size of the volume, in gigabytes
Snapshot snapshot.size gauge Size of the volume snapshot, in gigabytes
IP address ip.floating gauge Number of floating IP addresses

Cumulative metrics are polled every 5 minutes and increase over time, while gauge metrics are updated on events and show fluctuating values.

You can enable metering services in your compute cluster by doing one of the following:

  • If you have no compute cluster yet, deploy it and enable metering by adding the --enable-metering option to the vinfra service compute cluster create command. For example:

    # vinfra service compute create --nodes <node1_id>[,<node2_id>,...] --enable-metering
  • If you have already created the compute cluster, use the following command:

    # vinfra service compute cluster set --enable-metering


    Gnocchi will only take into account virtual machines created after metering has been enabled.

These commands open port 8041 and enable two Gnocchi services: gnocchi-api, an HTTP server, and gnocchi-metricd, a metric daemon.

After enabling metering, you can get access to your compute resource metrics either via REST API (see the official documentation) or using the Gnocchi command-line tool. To be able to use the tool, do as follows:

  1. Install the Gnocchi client:

    # yum install python-gnocchiclient
  2. Generate the admin OpenRC file:

    # su - vstoradmin -c 'kolla-ansible post-deploy'
    # source /etc/kolla/admin-openrc.sh

Now you can use the gnocchi command with the --insecure option. For example, to get a sum of vCPUs allocated to all virtual machines in the compute cluster and group them by project, use the following command:

# gnocchi --insecure measures aggregation --aggregation sum --metric vcpus \
--groupby=project_id --query=dummy --needed-overlap=0
| group                     | timestamp                 | granularity | value |
| project_id: 7decddff<...> | 2019-08-13T14:00:00+03:00 |       300.0 |   2.0 |
| project_id: 7decddff<...> | 2019-08-13T15:00:00+03:00 |       300.0 |   2.0 |
| project_id: b5bf7f3e<...> | 2019-08-13T14:00:00+03:00 |       300.0 |   4.0 |
| project_id: b5bf7f3e<...> | 2019-08-13T15:00:00+03:00 |       300.0 |   4.0 |

The output shows that at 2 and 3 p.m. on August 13, 2019 VMs included in the two projects were allocated 2 and 4 vCPUs, respectively.

To see the full list of gnocchi commands, refer to the official documentation.