Alerts
Perch can ping you the moment a health check goes down or a container crashes. Alerts come together in two steps: add a destination (where the notification goes), then create a rule (what to watch and when to fire).
Add a destination
A webhook for the service you want to notify, set up under Alerts in the sidebar.
Create a rule
Connect that destination to something worth watching, like a specific health check or container.
Destinations
A destination is just a webhook URL for the service you want notifications sent to. Perch supports three types:
Paste in your Discord webhook URL. Perch sends a rich embed with a colored sidebar (red for down or critical, green for recovery) and shows up as Perch Alerts with the Perch icon.
To grab a webhook URL in Discord, open your server settings, go to Integrations → Webhooks, and create a new webhook in whichever channel you like.
Paste in your Slack incoming webhook URL and Perch sends a plain message. To get one, head to your Slack workspace’s app directory and create an Incoming Webhooks app, or reuse an existing one.
Paste in your ntfy topic URL, like https://ntfy.sh/your-topic. Self-hosted ntfy works too. The notification lands on any device subscribed to that topic.
Rules
A rule connects a destination to something you want to keep an eye on. There’s no limit on rules, and more than one can point at the same destination.
Health check rules
Fires when a health check flips between up and down. It won’t fire on the very first result, so you don’t get spammed the instant you create a check. You’ll hear about it when the check goes down, and again when it recovers.
Container event rules
Fires on two kinds of container events:
- Crash: the container goes from
runningtostoppedordead - Restart: the container enters a
restartingstate
Pick the agent (host) and container you want to watch, or just watch every container on a host.
Cooldowns
Every rule has a cooldown. Once an alert fires, that rule stays quiet until the cooldown expires, which keeps a flapping check from flooding your notifications.
Tip
The cooldown resets after a successful recovery notification, so you’ll always hear about a fresh outage even if you were mid-cooldown.
Alert history
The bottom of the Alerts page has a collapsible history table of recent alerts. Each entry shows the rule, what triggered it, when it fired, and whether the notification went through or failed (a bad webhook URL, say).
Perch trims this history based on the Alert history retention setting in Instance Settings, which defaults to 90 days.