#Deploying

A deploy is when your project's code stops being a folder and becomes a live container with a public URL. CoDuck handles Docker image building, Postgres provisioning, migration runs, nginx + TLS, and starting the container. First deploy takes ~30–60 seconds; redeploys are ~10 seconds.

#Deploy from the dashboard

Open https://app.coduck.ai/project/<projectId>, switch to the Cloud panel → Hosting tab → Deploys sub-tab, and click Deploy. Watch progress in the Logs tab (tabbed alongside Hosting in the Cloud panel).

#What happens during a deploy

Every command comes from your coduck.json, which is re-read on every deploy — edit it, coduck push, coduck deploy, and the new commands run. Each phase streams to the Logs tab live:

1. Install — your install command, or npm install --include=dev --no-audit --no-fund by default.
2. Prisma generate — only if prisma/schema.prisma declares models.
3. Build — your build command, or npm run build by default. Set "build": "" to skip.
4. Schema bootstrap / pre-start — if you have a Prisma schema with models and no preStart, CoDuck runs prisma db push. If you set a preStart (e.g. prisma migrate deploy), that runs instead and you own schema setup.
5. Start — your start command, or npm start by default. nginx is then reconfigured to route traffic to the container.

The base image is Node 20 by default; Node 22 and static-site runtimes are available via runtime in coduck.json.

See the coduck.json reference for the full schema and field semantics.

#Instance size (resources)

Containers come in three tiers. Default is small; pick a bigger one with coduck deploy --size medium|large or "instanceSize" in coduck.json.

The Node build-heap ceiling auto-scales with the tier, so heavy client builds that OOM on small (three.js, remotion, recharts) usually succeed on large. A request above your plan is clamped down, and the deploy tells you.

Other limits: restart policy on-failure (2 retries), runs as non-root (uid 1001), working directory /app.

#Environment variables

Manage env vars in Cloud panel → Hosting → Env sub-tab. You can set, view (masked by default), update, and delete values. Changes take effect on the next deploy.

A set of keys are reserved — CoDuck injects them on every deploy and rejects them on write. They include DATABASE_URL, DIRECT_URL, PORT, NODE_ENV, and everything under CODUCK_* / NEXT_PUBLIC_CODUCK_*. The full, authoritative list is coduck env reserved; see the coduck.json reference. Stripe keys are reserved only once you connect Stripe.

coduck env import .env skips reserved/invalid keys (with a warning) and imports the rest — a .env containing DATABASE_URL won't abort the whole import.

#Stop, restart, tear down

In Cloud panel → Hosting → Deploys sub-tab you have three controls:

• Stop — stops the container. The database, URL, and env vars are preserved.
• Restart — restarts the container with the same config and a fresh runtime state.
• Destroy — removes the container, deletes the project's Postgres database, and removes the nginx vhost. Irreversible — take a backup first if you need the data.

#Custom domains and TLS

Once deployed, your project is reachable at <project>.coduck.app over HTTPS — nginx and a Let's Encrypt certificate are set up automatically on the first deploy. To attach your own domain, see Custom domains.

#Reading logs

Open Cloud panel → Logs tab, or from the CLI: coduck logs --follow, with --since 30m and --grep <pattern> to narrow down. coduck logs-digest summarizes errors/warnings.

#When a deploy fails

A failed deploy reports the specific cause (not a generic crash):

• "out of memory …" — the build or app exceeded the instance's RAM. Retry on a bigger tier: coduck deploy --size large (or medium). Moving asset compilation / codegen into the build step also helps.
• "the build failed …" — fix the error shown in the logs and redeploy. A common cause is a build-time package living in devDependencies — make sure your install step includes dev deps (the default does).
• "responding on :3000 but health-checking : …" — your app hardcodes a port. Listen on process.env.PORT instead.
• "the app crashed after starting …" — the start command exited; check coduck logs. Usually a missing env var or a failed DB connection (remember the database starts empty — run migrations in preStart).

#For developers

There's also a CLI (@coduckai/cli) for deploying from your terminal or from CI. See CLI commands.

#Next

• Custom domains — attach example.com to your project.
• Database overview — what DATABASE_URL points at, and how to work with it.