First alert in five minutes
From docker compose up -d to a message on your phone, without reading the rest of
the documentation. Each step says what you should see.
| Step | What you do | Typical time |
|---|---|---|
| 1. Start | docker compose up -d (first pull of the image) | 30 to 60 s |
| 2. Setup code | docker compose logs dumbmonit | 15 s |
| 3. Account | Open the page, accept the certificate once, create the admin account | 1 min |
| 4. Something to watch | Type one address, press Enter | 15 s |
| 5. Where alerts go | One click on ntfy (or Discord, Telegram, email, webhook), then Add channel | 30 s |
| 6. Proof | Send a test alert: the message arrives | 15 s |
About three and a half minutes, most of it the image download and the account.
1. Start the container
Section titled “1. Start the container”mkdir dumbmonit && cd dumbmonitcurl -fsSLO https://raw.githubusercontent.com/selfhostr/dumbmonit/main/docker-compose.ymldocker compose up -ddocker compose logs dumbmonit # the one-time setup code, for example K7XQ4-M9PRT2. Create the account
Section titled “2. Create the account”Open https://<your-host>:8080. The certificate is self-signed at first start: accept
the warning once (replace it later). Enter the setup code, a user name and
a password of at least 12 characters, then Create account and continue.
3. Your first alert
Section titled “3. Your first alert”The setup wizard opens on Your first alert, three links of one chain. The dot of each link turns green, with the word done, once it is proven.
-
What should we watch? Type one address and press Watch it:
https://nas.lanor any URL: a website check (HTTP status, certificate);192.168.1.10orrouter.lan: a ping;nas.lan:22: a TCP port.
The monitor is probed at once, not at the next minute: within a few seconds you read nas.lan answers. (or why it does not). Anything else (SNMP, Proxmox, the agent, a NAS) is under Another kind, in a new tab.
-
Where should the alert reach you? One click on a channel kind. The form only asks for what that kind needs, and the name defaults to the kind.
- ntfy is the quickest to a phone: a private topic is already filled in. Add the channel, install the ntfy app on your phone and subscribe to the topic shown, or open the link under it.
- Discord takes a webhook URL, Telegram a bot token and a chat ID, email an SMTP server, webhook any URL. Other channels lists all of them.
-
Prove it. Send a test alert sends a message to that channel through the same notifier as a real alert. You should read Test sent to ntfy and see the message on the phone.
That is the whole road. Continue goes on to the name and language of the instance, more monitors and the agent; Skip setup leaves the wizard at any time.
When does a real alert arrive?
Section titled “When does a real alert arrive?”The built-in rules are on from the start and notify every enabled channel. With the default 60 s check interval:
| The monitor | Rule | From the outage to the message |
|---|---|---|
| Website, ping or TCP port stops answering | Service down: two minutes of failed checks (at least three in a row) | 2.5 to 4 min |
| Any other monitor stops reporting | Monitor unreachable: no measurement for three minutes, confirmed on the next cycle | 3 to 4.5 min |
A single missed check never raises an alert. Alerts that fire close together leave as one message (30 s grouping window, Settings → Notifications). Every delay is a setting: the check interval on the monitor, the delay of each rule in Alerts → Rules.
Measuring it on your instance
Section titled “Measuring it on your instance”Settings → System → About → First run shows when this instance passed each step: started without an account, administrator created, first monitor, first channel, first message delivered (a test counts), first real alert delivered, each with its delay from the start. These times stay on the instance; nothing is sent anywhere.
- Add more monitors: SNMP, a network scan, Proxmox, the agent.
- Notification channels and alert rules.
- Back up Settings → Backup and
/data/secret.key: see Backup.