For the complete documentation index, see llms.txt. This page is also available as Markdown.

Alerts

Get an email when a device goes offline, a reading crosses a threshold, or a data source stops reporting. Alerts are evaluated in near real time, as your data streams into the Ardexa cloud.

Menu → Alerts

Alerts watch your devices and their data for you, and email the people you choose when something needs attention. Each alert is a condition, such as "an inverter has been offline for more than 10 minutes" or "an inverter's temperature has been above 75 °C for 15 minutes". When the condition is met, Ardexa raises an incident and notifies the alert's recipients.

The Alerts page, with three alerts selected and their incident history shown
The Alerts page. Alerts with an open incident are shown in red.

Alert types

Type
Raises an incident when…
Typical use

Device offline

a device has lost its connection to the Ardexa cloud, either immediately or after a time you choose (10 seconds to 1 day)

A site has lost power or communications

Query

readings from your sources meet a set of conditions, optionally for a minimum time and only during certain hours of the day

A temperature is too high, output is too low, a fault flag is set

Data is absent

a source has not sent data to a table for a time you choose, while other sources on the same device are still reporting

A logger or meter has stopped reporting

Data is absent alerts only fire while the device itself is still sending data. If the whole device goes quiet, use a Device offline alert instead. While a device is offline, the Ardexa agent keeps its data and sends it when it reconnects, so a short outage does not leave gaps.

How to set up each type is covered in Managing alerts.

How alerts and incidents work

  • Alert: the condition you set up, who it applies to, and who is notified.

  • Incident: a record that an alert's condition was met. One alert can have many incidents, one for each device or source that meets the condition. For example, if two inverters overheat, you get two incidents.

  • Active incident: an incident whose condition is still true. When the condition clears, the incident closes and a Resolved email is sent.

  • Acknowledged incident: an active incident that you have marked as seen. Acknowledging does not close an incident. It stays open until the condition clears.

  • Incident logs: every incident opening and closing is recorded permanently in the incident_logs table, even after the alert is changed or deleted. See Incident Logs.

Conditions with a duration are timed using the timestamps on your data, not the time the data reached the cloud. If a device sends readings late, for example after a communications outage, Ardexa works out when the condition really started and ended. If late data shows that a condition was never met, the incident is withdrawn and recipients get a Retraction email.

Notifications

Each alert has a list of email addresses to notify. You can pick workgroup members or type in any address. See Managing alerts.

Recipients get an email when:

  • Triggered: an incident opens.

  • Resolved: the condition has cleared and the incident has closed.

  • Retraction: late data showed the condition was never met, so the incident was withdrawn.

To be told only about new incidents, turn on Only send NEW notifications on the alert. Recipients then get the Triggered email and nothing else.

The email subject shows the status, workgroup, alert, device and severity. The body includes the alert's description, the device and source, its categories, and the time the incident started (or ended, for Resolved), shown in the device's time zone (or UTC if the device doesn't have one).

An example Triggered alert email
A Triggered email. The unsubscribe link is at the bottom.

To protect your inbox during an event that affects many devices at once, Ardexa limits how many alert emails it sends: currently 20 an hour for any one alert source, and 2,000 recipients an hour for a whole workgroup. Incidents are still recorded while email is paused, and workgroup owners get a notice explaining what was paused.

Unsubscribing from an alert

Every alert email has an unsubscribe link at the bottom. To stop receiving emails for that alert:

  1. Click unsubscribe in the email. You don't need to log in.

  2. Check the alert, workgroup and email address shown, then click Confirm Unsubscribe.

The Unsubscribe from Alert confirmation page
Confirming the unsubscribe

Your address is removed from that one alert's notification list. You still get emails from any other alerts you are on. If you change your mind, ask your workgroup owner, or anyone who can edit the alert, to add your address back.

Unsubscribe links expire after 30 days. If a link has expired, use the link in a more recent email from the same alert.

Who can use alerts

To…
You need

View alerts and incidents, and acknowledge incidents

Read access to the workgroup. Dashboard-only users can't open the Alerts page.

Create, edit and delete alerts, and manage categories

The Manage Alerts, Searches & Dashboards permission (or the Owner, Admin, Power or Standard role)

If you belong to a device group:

  1. You only see alerts set up for your device group.

  2. You can only create, edit and delete alerts for your own device group.

  3. You can use existing source labels in your alerts, but you can't change labelling settings.

On the Unified Data – Foundation Package, Query alerts and alert categories are not available. They are part of the Unified Data – Enterprise Package.

Before you start

  • Device groups: an alert can cover all devices in the workgroup, one device group, or a single device. Device groups are set up under Admin → Workgroup → Device Groups.

  • Source labels: Query and Data is absent alerts apply to the sources that have a chosen source label. Labels are set up under Admin → Workgroup → Labelling.

In this section

  • Managing alerts: create, edit and delete alerts, and set up categories.

  • Viewing incidents: see incident history, the incidents that are active now, and charts; acknowledge incidents.

  • Incident Logs: search the permanent record of every incident.

Last updated

Was this helpful?