Alerts and notifications

#Nothing is sent until you ask

A fresh install has no alert rules. A failed deploy is marked failed in the dashboard with its reason, and the old version keeps serving, but nobody is told until you add a rule or turn on notifications below.

#Alert rules

Create rules on the Alerts page. Each rule watches one app, every app, the host, or a service of Dockify itself.

For each rule you set the threshold, a duration from 1 to 1440 minutes, and the channels.

• Gauge rules (down, memory, disk, service) fire after the condition has held for the whole duration, and resolve after it has been clear for a minute. Memory and disk clear only 5 points under the threshold, so a value hovering at the line cannot flap.
• Window rules (crashLoop) already measure over the duration, so they fire at once.
• No data holds state. A tick where the sampler saw nothing never opens or resolves an alert.

There is one open alert per rule and app, so a flapping app does not send a stream of messages.

#Channels

A rule sends to any of these:

Add webhooks under Settings > Integrations > Webhooks, and send a test from there. Every alert carries a link to the page that shows the problem, such as the app's logs for a crash loop.

#The webhook body

A webhook receives one JSON object per event:

{
  "version": 1,
  "id": "0f0e0d0c-0000-4000-8000-0000000000aa",
  "event": "alert.firing",
  "occurredAt": "2026-11-06T09:30:00.000Z",
  "severity": "critical",
  "title": "web is down",
  "text": "No running container for 5 minutes",
  "url": "https://dockify.example.com/apps/...",
  "fields": [{ "name": "App", "value": "shop / web" }],
  "instance": { "name": "Acme Cloud", "url": "https://dockify.example.com" }
}

event is one of alert.firing, alert.resolved, update.available, update.updated, update.failed, update.rolled_back, deploy.failed, backup.failed, job.failed, tls.expiring and test. id stays the same across retries of one delivery. An alert also carries alert, app and service, and an update notice carries update.

Give a generic webhook a secret and each request is signed: X-Dockify-Signature is sha256= and the hex HMAC-SHA256 of <X-Dockify-Timestamp>.<body>. X-Dockify-Event and X-Dockify-Delivery come with every request.

Discord, Slack, Microsoft Teams, Google Chat, Telegram, Mattermost, Rocket.Chat and ntfy each get their own format, detected from the URL. Anything else is the generic webhook above.

#Mutes

Mute an app for 1, 8 or 24 hours during maintenance. Its alerts are still recorded, marked muted, but are not sent to webhooks or email.

#Instance notifications

Settings > Notifications turns on events for the whole instance: Deploy failed (webhooks, and email to whoever started it), App down (off, alert rules only show in the bell), Backup failed (webhooks, and email to admins), Scheduled job failed (the same, once when a job starts failing, until it succeeds again), TLS expiring (the same, for a certificate with less than 14 days left whose renewal keeps failing) and Preview deployments (email to whoever opened the pull request, or to admins). Only an admin can change them.

#Your own email

With email set up, Settings > Preferences has two choices for your own inbox:

• Deploys I started that fail. A deploy you started did not go live.
• Security alerts about my account. A new sign-in, or a password or two-factor change. Always on for admins.

#What can go wrong

• A disk alert, but cleanup freed little. Look at the disk breakdown on the Server page. Volumes and database data are never cleaned up for you.
• The webhook or Discord channel shows an error. Its last send did not get a 2xx answer. Send a test from Settings > Integrations to see what it answers.