Skip to main content
Alerting & Cases

Get Alerted When a Device Goes Offline

Learn how octoja detects that a device is offline and how to set up the Device online check so the right people are notified.

For servers, gateways, and other critical devices you usually want a notification the moment they stop responding. In octoja that is the job of the Device online check combined with an alert configuration. This article shows the setup and explains the timing behind the online status, so you can predict when a notification actually goes out.

How octoja decides a device is offline

Every agent keeps a live connection to your octoja instance and sends a heartbeat roughly every 30 seconds. A device counts as online while that connection is up and a heartbeat arrived within the last one and a half minutes. A device counts as offline once the heartbeat stays away longer than that.

Two short grace windows prevent false alarms:

  • After a brief disconnect (for example a router reconnect), the device has about half a minute to re-establish its connection before an offline verdict is recorded.
  • After octoja itself restarts (for example during maintenance), devices get a few minutes to reconnect before any offline verdicts are recorded, so a platform restart does not page you for your whole fleet.

For network devices probed by a monitoring device, there is one exception: while the monitoring device itself is offline, the Device online check does not evaluate those network devices and records no result for them. This way a failed monitoring device does not raise a separate alert for every network device behind it. Monitor the monitoring device itself with Device online as well.

Set up the offline alert

  1. Go to Devices and open the device you want to watch.
  2. Click the Checks tab → Add Check → Device online, then click Add Check. The check has no thresholds to configure — it reports OK while the device is online and Critical while it is offline.

The Add Check dialog on ADVENTURE-WEB01 with the Device online check selected, showing the check summary and the Cancel and Add Check buttons at the bottom.

  1. Attach an alert configuration to the check — the walkthrough is in Set up alerts for a check. To roll the check out to many devices at once, assign it through a config package instead: Assign checks with config packages.

Tune how fast — and how often — you are notified

The alert configuration's window decides how many failing results it takes before a channel fires. That is your tuning knob between "know immediately" and "no noise from one-minute blips":

GoalRule of thumb
React on the first offline resultEvaluate 100% of the last 1 result at minimum severity Critical. The connection grace period still applies; the channel must be active
Require sustained failureUse a larger result window and a suitable percentage. Requiring two critical results can add another five-minute check interval before notification

One thing octoja never does: repeat an open alarm at intervals. Each channel notifies once when its rule first crosses the threshold. It notifies again on recovery — except SMS and Call, which notify you of the alarm only. If you receive bursts of offline-and-recovered messages for many devices at the same moment, that points at a shared network path (site router, firewall, ISP) briefly dropping — the per-device grace windows absorb single-device blips, not site-wide ones. The full notification semantics are described in Set up alerts for a check.

Tips

  • Reserve call- or SMS-based channels for a small window on genuinely critical devices, and use e-mail or case channels with a larger window for everything else.
  • The device's current online state and the last-seen timestamp are always visible on the Devices page — see Device Statuses.
  • If many devices across different customers report offline at exactly the same time and recover seconds later, check octoja Availability & Status before suspecting the devices.

FAQ

If the device already had a Device online check, open its check history to review the recorded results. This is not a complete connection log: brief interruptions within the grace period do not produce an offline result, and adding the check later does not reconstruct earlier outages.

For a closer investigation, collect the local agent logs and give support the approximate time of the interruption, including the time zone.

SMS and Call alarm groups notify you of the alarm only, not the recovery. There is no additional group switch to enable this. To receive recovery notifications as well, add another channel to the same alert configuration, for example Email. See Set up alerts for a check for setup.