Docs

Appearance

Open console

Administration

Monitoring

Watch the health and load of your Thinger.io instance with Cluster Hosts, the health check endpoint and the Prometheus Exporter plugin.

Guide1 min read

This guide shows how to keep an eye on your Thinger.io instance: check its load in the console, alert when it stops answering, and export its metrics to your own monitoring tools.

Your instance/v1/server/healthcheckno token neededCluster HostsOverview and LogsPrometheus Exporterplugin, metrics endpointAlarmsrules on your devicesUptime monitorcalls it periodicallyAdministratorchecks load and outputPrometheusscrapes the metricsEndpointsemail, webhooks, chatAlertwhen it fails or is slowGrafana, Alertmanagerdashboards and alertsDevice alertoffline or out of range
Ways to monitor an instance

Server monitoring client discontinued

The Thinger.io server monitoring client (thinger_monitor) is no longer maintained: do not use it for new installations. Use the options on this page instead. To monitor and manage Linux machines as devices, see ThinRemote.

Before you start ​

  • An administrator account on a private instance. YOUR_SERVER is the hostname of your instance.
  • For external alerts, an uptime monitoring service or tool that can call a URL.

Steps ​

  1. Check the load of the server

    Go to Administration > Cluster Hosts, click the host and open Overview. It shows the current connections and a week of CPU, memory, disk and connection averages. The Logs tab follows the server output live. See what each counter means and how to size the server.

    Overview tab of a cluster host with the connection counters, and the charts and gauges of connections, CPU, RAM and disk, still without data
    Overview tab of a host

  2. Alert when the server stops answering

    Point your uptime monitor at the health check of the instance. It needs no token and answers 200 OK while the server works:

    bash
    curl -i "https://YOUR_SERVER/v1/server/healthcheck"

    Set the monitor to alert you when the call fails or takes too long.

  3. Export metrics to Prometheus

    Install the Prometheus Exporter plugin from Plugins > Marketplace. It publishes metrics about your devices and other resources in the Prometheus format, so that Prometheus, Grafana or Alertmanager can collect them and alert on them. The plugins of the Monitoring category cover the rest of the stack.

  4. Get notified about devices

    To know when a device disconnects or a reading goes out of range, create an alarm instead of watching the server.

Result ​

The console shows the load of every host, your uptime monitor alerts you if the instance stops answering, and your monitoring stack collects the metrics of your resources.

Next steps ​

Cluster HostsWatch the connections, CPU, memory and disk of each server of your instance, follow its logs live and restart it.Prometheus ExporterExport Thinger's resource status in a Prometheus formatAlarmsDefine alarm rules that watch bucket data, device properties or connection status, and notify through endpoints when a condition is met.