Skip to content

Create your first offline alert #

Finish with one test device monitored for sustained connectivity loss, an alert when it stays offline, and a recovery result after it reconnects. This checks Dataplicity connectivity; application health uses a different monitor.

Before you start #

Use a connected staging device that you can disconnect and restore locally, and a role that can create device monitors. Keep the device in your engineering organisation for this exercise. Each device can have one connectivity monitor. Do not interrupt a production device to test an alert.

1. Monitor the device #

  1. Open Device monitors and choose Monitor device.
  2. Search by the device's name, serial or ID and select the intended unit.
  3. Confirm its identity and online state in the form.
  4. Choose P2 for this test, then choose Start monitoring.
  5. Wait for the first evaluation and inspect Recent checks.

Expected result: monitoring is active and the online device passes its check. If it is already monitored, inspect its existing monitor instead of creating a duplicate.

2. Verify the offline alert #

  1. Using local access, disconnect the test device's network. Record the time.
  2. Wait for the monitor's consecutive failed evaluations to trigger.
  3. Inspect the monitor and the resulting alert. Confirm they identify the same test device.

A brief disconnection may not trigger an alert: consecutive checks suppress short connection flaps. Use the saved monitor's effective evaluation settings rather than expecting an immediate alert.

3. Verify recovery #

Restore the device's network locally. Wait for it to reconnect and pass the required recovery evaluations. Inspect the monitor's Connectivity and Recent checks history. It may enter cooldown before it can alert again.

Check it worked #

You are finished when the history shows online, sustained offline and recovered states for the intended device, with the corresponding alert. This proves the connectivity signal; verify the customer's actual application separately.

If it does not work #

SymptomNext check
Device cannot be selectedCheck its organisation, role access and whether it already has a monitor
No alertCheck that monitoring is active and enough failed evaluations have elapsed
No recoveryConfirm the original device record reconnects, then wait for successful evaluations
You need paging or escalationConfigure and test the supported incident and on-call path separately

Next task #

Use Device offline for a real connectivity incident. For a cohort, continue with Device monitors to choose tag selectors, offline counts or ratios.

Choose another monitor type #

You need to watchUse
HTTP endpoints, CORS, DNS or heartbeatsService monitors
Application-produced journey or metric healthUser-impact monitors
Matching log events sent to a webhookLog alert rules

Process, disk, TCP socket and arbitrary script checks are not current monitor types. Service, heartbeat and user-impact signals can feed incident automation and on-call.