1
0
mirror of https://github.com/jcwimer/openstack-exporter synced 2026-03-24 17:44:42 +00:00

openstack_exporter

This is a prometheus exporter for Openstack

Usage

With Docker

This script will forward all "OS_" variables to the container and start the container on port 8000

bash run-with-docker.sh --cloud_name openstack

The docker image includes Chromium for getting Horizon metrics.

With Python

Need Python 3.8 and pipenv

pipenv sync
pipenv run python openstack_exporter.py --cloud_name openstack

Optional Params

To deploy an instance and test the time to ping on all hypervisors, use the following flags:

--image IMAGE_NAME_OR_ID --flavor FLAVOR_NAME_OR_ID --network NETWORK_NAME_OR_ID --instance_deploy

The network used needs to have TCP port 22 (uses TCP instead of ICMP to ping) open in the default security group.

To capture horizon login response time, use the following flag:

--horizon_url "http://url_here"

This will use selenium to log into horizon. You will need Chrome or Chromium installed to use this feature. You will see these metrics:

openstack_api_response_seconds{api_name="horizon",cloud_name="CLOUD_NAME"}
openstack_api_status{api_name="horizon",cloud_name="CLOUD_NAME"}

Information

Standard Metrics Provided

Metric Metric Labels Description
openstack_api_response_milliseconds {api_name="API_NAME",cloud_name="CLOUD_NAME"} Milliseconds for the api to respond via openstack sdk. nova, neutron, and cinder are currently recorded.
openstack_api_status {api_name="API_NAME",cloud_name="CLOUD_NAME"} Status of the openstack api. 1 = up 0 = down. nova, neutron, and cinder are currently recorded.
openstack_hypervisor_running_vms {hypervisor_hostname="HYPERVISOR_NAME",cloud_name="CLOUD_NAME",aggregate="AGGREGATE_NAME"} Number of running VMs on every hypervisor in the region.
openstack_hypervisor_used_ram_mb {hypervisor_hostname="HYPERVISOR_NAME",cloud_name="CLOUD_NAME",aggregate="AGGREGATE_NAME"} Amount of RAM in MB used (as reported by nova-compute) for every hypervisor in the region.
openstack_hypervisor_total_ram_mb {hypervisor_hostname="HYPERVISOR_NAME",cloud_name="CLOUD_NAME",aggregate="AGGREGATE_NAME"} Amount of RAM in MB in total (as reported by nova-compute) for every hypervisor in the region.
openstack_hypervisor_used_cpus {hypervisor_hostname="HYPERVISOR_NAME",cloud_name="CLOUD_NAME",aggregate="AGGREGATE_NAME"} Number of vcpus used (as reported by nova-compute) for every hypervisor in the region.
openstack_hypervisor_total_cpus {hypervisor_hostname="HYPERVISOR_NAME",cloud_name="CLOUD_NAME",aggregate="AGGREGATE_NAME"} Number of vcpus in total (as reported by nova-compute) for every hypervisor in the region.
openstack_hypervisor_enabled {hypervisor_hostname="HYPERVISOR_NAME",cloud_name="CLOUD_NAME",aggregate="AGGREGATE_NAME"} nova-compute status for every hypervisor in the region. 1 = enabled 0 = disabled
openstack_hypervisor_up {hypervisor_hostname="HYPERVISOR_NAME",cloud_name="CLOUD_NAME",aggregate="AGGREGATE_NAME"} nova-compute state for every hypervisor in the region. 1 = up 0 = down
openstack_hypervisor_local_gb_total {hypervisor_hostname="HYPERVISOR_NAME",cloud_name="CLOUD_NAME",aggregate="AGGREGATE_NAME"} Total local disk in GB (as reported by nova-compute) for every hypervisor in the region.
openstack_hypervisor_local_gb_used {hypervisor_hostname="HYPERVISOR_NAME",cloud_name="CLOUD_NAME",aggregate="AGGREGATE_NAME"} Total local disk used in GB (as reported by nova-compute) for every hypervisor in the region.

Optional Metrics (use flags when running)

Metric Metrics Labels Description
openstack_instance_deploy_seconds_to_ping {hypervisor_hostname="HYPERVISOR_NAME",cloud_name="CLOUD_NAME"} Seconds from deploy command to ping when creating an instance for every hypervisor in the region. Requires --flavor, --image, --network, and --instance_deploy flags. The network used needs to have TCP port 22 (uses TCP instead of ICMP to ping) open in the default security group.
openstack_horizon_response_seconds {cloud_name="CLOUD_NAME"} Seconds it takes for Chromium to log into Horizon. Requires --horizon_url flag.
openstack_horizon_status {cloud_name="CLOUD_NAME"} Horizon status. 1 = up 0 = down. Requires --horizon_url flag.
Description
No description provided
Readme MIT 165 KiB
Languages
Python 92%
Dockerfile 6%
Shell 2%