This page takes a fresh server to a working Marmot with one monitor, one notification channel and one public status page. It assumes Docker; Development covers running from source and Deployment the production details (reverse proxies, backups, upgrades).

Requirements: Docker Engine 24+ with the compose plugin, and, for HTTPS, a DNS name that points at the host with ports 80 and 443 reachable from the internet.

mkdir marmot && cd marmot
base=https://raw.githubusercontent.com/ThinkHumanDotDev/marmot/main/docker
curl -fsSL $base/docker-compose.yml -o docker-compose.yml
curl -fsSL $base/Caddyfile -o Caddyfile
curl -fsSL $base/.env.example -o .env

Open .env and set at least:

Then:

docker compose up -d --wait
docker compose logs -f web      # watch the migrations run and the server start

The stack is five containers: web (UI, API, status pages), worker (runs the checks), realtime (live updates), postgres and redis, with caddy in front. web applies database migrations on every start, so there is nothing to run by hand. To use MongoDB instead of Postgres, download docker-compose.mongo.yml from the same place and add -f docker-compose.yml -f docker-compose.mongo.yml to every docker compose command.

Open https://status.example.com (or http://localhost). While the instance has no users every page redirects to /setup, which asks for:

• your name, email and a password (8 characters or more);
• the name and URL slug of your first organization (the slug becomes part of every URL, e.g. /acme/monitors).

Submitting creates the instance superadmin and the organization with you as its owner in one transaction and signs you in. The wizard closes for good after that: later visitors land on /login, and new people join through /signup (while sign-up is allowed) or an invitation (Organizations and members).

Superadmins can additionally open the Payload admin panel at /admin, which exposes Instance settings (public URL, sign-up, retention, proxy trust) and raw access to every collection. Day-to-day work happens in the Marmot UI, not the admin panel.

Go to Monitors → New monitor. The form asks for:

1. Type. Start with HTTP(s): UP when the response status is accepted (200-299 by default). Other types (keyword, JSON query, TCP port, ping, DNS, push, group, manual, and the extended set landing in the current release) are described in Monitors.
2. Name and URL (or hostname/port for the host-based types).
3. Timing: check every interval seconds (default 60, minimum 20), retry maxRetries times at retryInterval before going DOWN (default 0 retries), and resendInterval to re-alert while a monitor stays down (default 0 = alert once per transition).

Notification channels marked default are attached automatically; pick other channels per monitor in the form. Save. The worker schedules the monitor immediately and the first heartbeat shows up on the detail page within one interval, together with the uptime cards (24h/30d), the response-time chart and the list of status changes. The monitor list on /acme/monitors updates live over the WebSocket connection.

Go to Notifications → New channel, pick a provider (Slack, Discord, Telegram, SMTP email, ntfy, Gotify, Pushover, webhook, PagerDuty, …; the full list is in Notifications) and fill in the form that the provider describes. Send test delivers a test message before you save. Tick Default to attach the channel to every monitor created afterwards and Apply to all existing monitors to attach it to the monitors you already have (for now this is how channels reach monitors; picking channels per monitor in the monitor form lands in the current release).

Marmot alerts on status transitions (UP→DOWN, DOWN→UP, PENDING→DOWN) and, when resendInterval is set, every N consecutive DOWN beats.

Go to Status pages → New page, give it a title and a slug (/status/<slug>), then on the Groups & monitors tab create a group ("Core services") and drag monitors into it. Switch Published on in the header. The page is now public at https://status.example.com/status/<slug>, shows the last 50 beats and the 24h/30d uptime of each monitor, and refreshes itself every few minutes. Post incidents from the Incidents tab; the Domains tab serves the page on its own hostname (status.yourproduct.com), see Status pages.

Members → Invite sends an email invitation with a role (viewer, member, admin or owner), or copy the organization's invite link for a self-service join. With SMTP unset, invitation emails are printed to the web container log (docker compose logs web) so you can copy the link from there. Configure SMTP_* in .env to send real mail, and OIDC_* to let people sign in with your identity provider (Single sign-on).

• Harden and operate the install: Deployment (reverse proxies, backups, upgrades).
• Tune it: Configuration (every variable), the instance settings in /admin.
• Automate it: badges, the push endpoint, Prometheus metrics and API keys in Integrations.