Skip to main content
An alert source, or connector, turns alerts from another monitoring tool into Squasher incidents. A firing alert opens an incident, and the recovery resolves the same incident. The incident then goes through your notification routes, like an incident from a Squasher monitor.
The dashboard has no screen for alert sources yet. Create and manage them with the Squasher CLI or the Alert Connectors API.

​
Supported sources

​
Create a connector

You need write access to the project. The steps are the same for every source; only the --provider value and the setup in the source tool change.
1

Sign in to the CLI

2

Create the connector

Add --severity sev-2 to change the default severity. Without it, the default is sev-1.
3

Save the endpoint and secret

The command prints the connector and a setup object:
Squasher shows the secret only here and after a rotation. It stores only a hash. Save the secret in your secret manager.
4

Set up the source tool

Follow the page for your source in the table above. Send the secret in the Authorization: Bearer header, never in the URL.
5

Test with a real alert

Make a test alert fire, then recover. Check that one incident opens and that the same incident resolves.
squasher alert-connectors test <connector-id> does not send an alert. It reports whether the connector is enabled and when Squasher last received an event from it.

​
How alerts become incidents

Each source gives Squasher two values. The alert key stays the same from the first firing alert to the recovery, so every update lands on one incident. The event key is different for each delivery, so a retried delivery does not change the incident twice. After an incident resolves, the next firing alert with the same alert key opens a new incident. Squasher ignores an event that is older than the last event it applied for the same alert key. Incidents from connectors have the source type integration. You can use it to scope a notification route to alerts from other tools.

​
Severity

Squasher reads the severity from the alert and maps common names to incident severities:

​
Limits

​
Check deliveries

List the last deliveries and their results:
Each delivery gets one of these responses: If an incident opens but never resolves, check that the source sends recovery notifications and that the alert key does not change. Never use a timestamp or a retry ID as the alert key: every delivery then opens a new incident.

​
Rotate, disable, or delete a connector

  • Rotate the secret: squasher alert-connectors rotate-secret <connector-id>. The old secret works for 24 hours more. Update the source tool in that time. You cannot rotate again until the 24 hours end. CloudWatch connectors have no secret to rotate.
  • Disable: squasher alert-connectors update <connector-id> --enabled false. Squasher then answers the endpoint with 404. Open incidents stay open.
  • Delete: squasher alert-connectors delete <connector-id>. Squasher removes the secret and resolves every open incident from this connector. You can still read its delivery history.
Add --project "$SQUASHER_PROJECT_ID" to each command, or set SQUASHER_PROJECT_ID in your shell.

​
Next steps

Route incidents

Send incidents from alert sources to Slack, PagerDuty, email, Telegram, or a webhook.

Alert Connectors API

Create and manage connectors over HTTP.