Monitor Sensu with Sensu

This guide describes best practices and strategies for monitoring the Sensu backend with another Sensu backend or cluster.

To completely monitor Sensu (a Sensu backend with internal etcd and an agent), you will need at least one independent Sensu instance in addition to the primary instance you want to monitor. The second Sensu instance will ensure that you are notified when the primary is down and vice versa.

This guide requires Sensu plugins using dynamic runtime assets. For more information about using Sensu plugins, see Use dynamic runtime assets to install plugins.

NOTE: This guide describes approaches for monitoring a single backend. These strategies are also useful for monitoring individual members of a backend cluster.

This guide does not describe Sensu agent keepalive monitoring.

The following ports and endpoints are monitored as part of this guide:

Port Endpoint Description
2379 /health Etcd health endpoint. Provides health status for etcd nodes.
8080 /health Sensu Go health endpoint. Provides health status for Sensu backends, as well as for Postgres (when enabled).

Monitor your Sensu backend instances

Monitor the host running the sensu-backend locally by a sensu-agent process for operating system checks and metrics.

For Sensu components that must be running for Sensu to create events, you should also monitor the sensu-backend remotely from an independent Sensu instance. This will allow you to monitor whether your Sensu event pipeline is working.

To do this, use the check_http plugin from the Monitoring plugins dynamic runtime asset to query Sensu’s health API endpoint with a check definition for your primary (Backend Alpha) and secondary (Backend Beta) backends:

type: CheckConfig
api_version: core/v2
metadata:
  namespace: default
  name: check_beta_backend_health
spec:
  command: check-http.rb -u http://sensu-backend-beta:8080/health -n false
  subscriptions:
    - backend_alpha
  interval: 10
  publish: true
  timeout: 10
  runtime_assets:
    - sensu-ruby-runtime
    - sensu-plugins-http
type: CheckConfig
api_version: core/v2
metadata:
  namespace: default
  name: check_alpha_backend_health
spec:
  command: check-http.rb -u http://sensu-backend-beta:8080/health -n false
  subscriptions:
    - backend_beta
  interval: 10
  publish: true
  timeout: 10
  runtime_assets:
    - sensu-ruby-runtime
    - sensu-plugins-http

Monitor external etcd

If your Sensu Go deployment uses an external etcd cluster, you’ll need to check the health of the respective etcd instances.

This example includes checks for your primary (Backend Alpha) and secondary (Backend Beta) backends:

type: CheckConfig
api_version: core/v2
metadata:
  namespace: default
  name: check_beta_etcd_health
spec:
  command: check_http -H sensu-beta-etcd -p 2379 -u /health
  subscriptions:
    - backend_alpha
  interval: 10
  publish: true
  timeout: 10
  runtime_assets:
    - monitoring-plugins
type: CheckConfig
api_version: core/v2
metadata:
  namespace: default
  name: check_alpha_etcd_health
spec:
  command: check_http -H sensu-alpha-etcd -p 2379 -u /health
  subscriptions:
    - backend_beta
  interval: 10
  publish: true
  timeout: 10
  runtime_assets:
    - monitoring-plugins

Monitor Postgres

COMMERCIAL FEATURE: Access enterprise-scale Postgres event storage in the packaged Sensu Go distribution. For more information, see Get started with commercial features.

Larger Sensu deployments may use Postgres as an alternative datastore to process larger numbers of events. The connection to Postgres is exposed on Sensu’s /health endpoint and will look like the example below:

{
	"Alarms": null,
	"ClusterHealth": [{
		"MemberID": 3470366781180380542,
		"MemberIDHex": "302938336092857e",
		"Name": "sensu00",
		"Err": "",
		"Healthy": true
	}, {
		"MemberID": 15883454222313069303,
		"MemberIDHex": "dc6d5d7607261af7",
		"Name": "sensu01",
		"Err": "",
		"Healthy": true
	}, {
		"MemberID": 11377294497886211005,
		"MemberIDHex": "9de44510fb838bbd",
		"Name": "sensu02",
		"Err": "",
		"Healthy": true
	}],
	"Header": {
		"cluster_id": 13239446193995634903,
		"member_id": 3470366781180380542,
		"raft_term": 1549
	},
	"PostgresHealth": [{
		"Name": "sensu_postgres",
		"Active": true,
		"Healthy": true
	}]
}

To monitor Postgres’ health from Sensu’s perspective, use a check like this example:

type: CheckConfig
api_version: core/v2
metadata:
  created_by: admin
  labels:
    sensu.io/managed_by: sensuctl
  name: check-postgres-health
  namespace: default
spec:
  check_hooks: null
  command: check-http-json.rb -u https://sensu.example.com:8080/health --key PostgresHealth[0].Healthy
    --value true
  env_vars: null
  handlers: []
  high_flap_threshold: 0
  interval: 10
  low_flap_threshold: 0
  output_metric_format: ""
  output_metric_handlers: null
  proxy_entity_name: ""
  publish: true
  round_robin: true
  runtime_assets:
  - sensu-plugins/sensu-plugins-http
  - sensu/sensu-ruby-runtime
  secrets: null
  stdin: false
  subdue: null
  subscriptions:
  - backends
  timeout: 0
  ttl: 0
{
  "command": "check-http-json.rb -u https://sensu.example.com:8080/health --key PostgresHealth[0].Healthy --value true",
  "handlers": [],
  "high_flap_threshold": 0,
  "interval": 10,
  "low_flap_threshold": 0,
  "publish": true,
  "runtime_assets": [
    "sensu-plugins/sensu-plugins-http",
    "sensu/sensu-ruby-runtime"
  ],
  "subscriptions": [
    "backends"
  ],
  "proxy_entity_name": "",
  "check_hooks": null,
  "stdin": false,
  "subdue": null,
  "ttl": 0,
  "timeout": 0,
  "round_robin": true,
  "output_metric_format": "",
  "output_metric_handlers": null,
  "env_vars": null,
  "metadata": {
    "name": "check-postgres-health",
    "namespace": "default",
    "labels": {
      "sensu.io/managed_by": "sensuctl"
    },
    "created_by": "admin"
  },
  "secrets": null
}

A successful check result will look like this:

Screenshot of a successful Postgres health check in Sensu Go web UI

Successful Postgres health check in Sensu Go web UI