dployr

Appearance

Health checks and Watchdog ​

Point dployr at a health check endpoint and Watchdog takes it from there. It polls your service, marks it degraded if it stops responding, and notifies you on whatever channels you have set up.

Health checks ​

A health check is an HTTP request dployr makes to your service on a regular interval. Any non-5xx response is treated as healthy, including 4xx. A 5xx or no response marks the service degraded.

Configuring a health check ​

Set the path in your blueprint:

yaml
health_check: /health

Or from the service's Settings tab in the dashboard. That's it. Watchdog handles the polling.

The endpoint just needs to respond without a 5xx. A dedicated /health route that returns 200 is the cleanest approach, but for better signal check the things your app actually depends on: database connectivity, cache availability, external services.

Watchdog ​

The watchdog is dployr's alerting layer. It runs on top of health checks and detects two things:

  • Services that fail health checks (degraded)
  • Services that stop responding entirely without crashing (hangs and deadlocks)

The second case is what makes it useful beyond basic process monitoring. A service can be running at the OS level but completely stuck. The watchdog catches this.

Notification channels ​

Configure where you want to be notified from the dashboard under Settings > Integrations.

Slack: paste your Slack webhook URL. Alerts fire into the channel you configured on the webhook.

Discord: paste your Discord webhook URL. Same setup as Slack.

Email: enter one or more email addresses. Alerts go out on first failure and on recovery.

Webhook: any URL. dployr will POST a JSON payload when a service goes degraded and again when it recovers:

json
{
  "event": "service.degraded",
  "service": "my-api",
  "cluster": "my-cluster",
  "timestamp": "2025-01-01T12:00:00Z",
  "message": "Health check failed"
}

You can point this at PagerDuty, Opsgenie, your own alerting pipeline, or anything that accepts a POST.

Recovery alerts ​

When a service recovers, dployr sends a follow-up notification on the same channels. You'll know when something went wrong and when it came back without having to check the dashboard.

Viewing health status ​

From the services list, each service shows its current health state next to its running status. From the service detail page, the Overview tab shows a health timeline so you can see when issues occurred and how long they lasted.