atlas

Atlas, brother of Prometheus.

This is a standalone custom Prometheus service discovery application. It consists of a growing number of discoveries which can be enabled/disabled and configured via a yaml file. The discovered targets are then written into a configmap file, which is consumed by Prometheus.

Architecture overview

Available Discoveries:

1. Ironic Nodes

discoveries:
      ironic:
        refresh_interval: 600 #How often the discovery should check for new/updated nodes.
        targets_file_name: "ironic.json" #Name of the file to write the nodes to.
        os_auth: # Openstack auth
          auth_url: openstack auth url
          user: openstack user
          password: os user pw
          user_domain_name: openstack user_domain_name
          project_name: openstack project_name
          domain_name: openstack domain_name

2. Netbox API

• DCIM-Devices

  netbox:
      refresh_interval: 600 # How often the discovery should check for new/updated devices.
      targets_file_name: "netbox.json"  #Name of the file to write the devices to.
      netbox_host: "netbox_host_url"
      netbox_api_token: "netbox_api_token"
      dcim:
        devices: #Array of device queries
          - custom_labels: #Use to add custom labels to the target
              anyLabel: "anyValue"
              job: "job_name"
            target: 1 #Query Parameters: Any parameters the netbox api ([netbox_url]/api/dcim/devices/) accepts.
            role: "role_name"
            manufacturer: "cisco"
            region: "de1"
            status: "1"
          - custom_labels: ....
  

• Virtualization-VMs

  netbox:
      refresh_interval: 600 # How often the discovery should check for new/updated devices.
      targets_file_name: "netbox.json"  #Name of the file to write the devices to.
      netbox_host: "netbox_host_url"
      netbox_api_token: "netbox_api_token"
      virtualization:
        vm: #Array of vms queries
          - custom_labels: #Use to add custom labels to the target
              anyLabel: "anyValue"
              job: "job_name"
            target: 1 #Query Parameters: Any parameters the netbox api ([netbox_url]/api/virtualization/virtual-machines/",) accepts.
            manufacturer: "cisco"
            region: "de1"
            tag: "tag_name"
          - custom_labels: ....
  

Install

A Dockerfile is provided to run it on Kubernetes. All necessary ENV VARs/flags can be figured out running ipmi_sd --help:

NAME:
   atlas - discovers custom services, enriches them with metadata labels and writes them to a file or Kubernetes configmap
 USAGE:
   atlas [global options]
 VERSION:
   0.1.7
 COMMANDS:
     help, h  Shows a list of commands or help for one command
 GLOBAL OPTIONS:
  - OS_PROM_CONFIGMAP_NAME: name of the configmap, where the discovered nodes should be written to.
  - K8S_NAMESPACE: name of the K8s namespace atlas is running in.
  - K8S_REGION: name of the k8s region atlas is running in
  - LOG_LEVEL: log level atlas should use:
    - "debug"
    - "error"
    - "warn"
    - "info"

To figure these out you can also just run it locally. Either by building via docker using docker build . and then docker run CONTAINER --help or directly on bare metal if you have a working go environment with go run cmd/atlas/main.go --help.

Prometheus server configuration: A sample prometheus job to read those configmap targets is shown here