Alerts
Alerts emails the right people when something needs attention. Each rule reads like a sentence: what to watch, how much is too much, and who hears about it.
Open Monitor › Alerts in the left sidebar.
Alert rules and their history
Alert emails never contain patient information. They name the feed, the destination or the field, and the numbers.
What a rule can watch for
| Watch for | Raises an alert when |
|---|---|
| Failure rate above threshold | Too large a share of the messages received in the window did not reach every destination. |
| Dead letters above threshold | The platform gives up on more messages than expected, which usually means a destination is refusing them. |
| Destination failing repeatedly | One destination keeps rejecting or timing out, even while the other interfaces look healthy. |
| No messages received | A feed that normally sends messages goes quiet, so silence is not mistaken for a quiet day. |
| Processing slower than threshold | Messages are taking longer than usual to get from arrival to delivery. |
| Messages rejected on a specific field | Destinations start refusing messages because of a particular field. This usually means a mapping is wrong or the sending system changed what it puts there. |
| Field anomaly: fields missing from a feed | Fields that arrived reliably go missing from new messages, for example after the sending system is upgraded. See Field watch. |
Create a rule
-
Select New rule.
-
Enter a Name. It is the subject line of the email, so make it recognisable at 3am, for example
Hospital PAS feed failing. -
Under Watch for, choose what to watch.
A new alert rule
-
Set how much is too much. The fields depend on what you watch, for example Percentage of messages failing and Looking back over.
-
Set Remind at most, for example Once per 1 hr. While the problem lasts, reminders are held back to this rate.
-
Optionally limit the rule to Only this endpoint or Only this destination. Leave them empty to cover everything.
-
Under Tell these people, enter email addresses separated by commas. A shared inbox usually survives staff changes better than an individual.
-
Leave Active on.
-
Select Test rule to measure it against real traffic now. It sends nothing.
-
Select Create rule.
Thresholds, recipients and the test
Rules are checked every few minutes.
Watch for field anomalies
- Select New rule.
- Name it, for example
Field anomaly. - Under Watch for, choose Field anomaly: fields missing from a feed.
Missing fields allowed is set to
0, so any missing field alerts. - Enter the recipients and select Create rule.
The email names each field, its feed and how often it arrived, for example "PID-3 (Patient identifier list) on simulated-adt-feed: was 100%, now 0%".
History
History lists everything the platform noticed, newest first:
- Triggered: the rule fired and the people were emailed.
- Suppressed: the problem was still there, so the reminder was held back.
- Resolved: the problem cleared, and a RESOLVED email was sent.
Everyone in the workspace can see rules and history. Only admins can create, change, test or remove rules.