Container catalyst-backend fails to start

[HesusAIM]

I'm trying to use One-Line Install and it fails. Version 1.15.13

[Dziondzio]

same, any fix?

[karutoil]

Hey!

Sorry for the wait on this, I had this project on hold while I was focusing on some life changing personal things!

Will do some testing on my end to ensure that the docker compose runs without issues.

Thanks,
Marcus!

[karutoil]

Root cause

Two separate packaging bugs were hiding behind the one-liner install crash (v1.15.13 image):

1. Cannot find package 'dotenv' — with node-linker=hoisted, the backend Docker runtime stage only copied catalyst-backend/node_modules. Those entries are relative symlinks into the monorepo-root .pnpm store (../../node_modules/.pnpm/...). In the final image that path resolves outside /app, so every production dependency (dotenv, fastify, …) was a broken symlink.

2. FATAL: database "catalyst" does not exist — the Postgres healthcheck ran pg_isready -U catalyst with no -d, so libpq defaulted to a DB named after the user (catalyst). The real DB is catalyst_db. That produced the spam in the screenshot; it was noisy, not the boot-stopper.

Once dotenv was fixed, a third issue surfaced: tsc emits extensionless ESM imports (from "./db") which plain Node cannot resolve. Migrations also need the Prisma CLI in the image.

Fix (landed on main)

• Backend image now uses pnpm deploy --prod --legacy for a self-contained node_modules
• prisma moved to production dependencies so migrate deploy works on boot
• Post-tsc pass rewrites relative imports to .js
• Postgres healthcheck uses pg_isready -U catalyst -d catalyst_db
• Entrypoint prefers the local Prisma CLI over npx

Verified locally: image builds, migrations apply, /health returns {"status":"ok"}.

What you need to do

After CI publishes a new ghcr.io/catalystctl/catalyst-backend:latest:

cd catalyst-docker
docker compose pull
docker compose up -d
docker compose logs -f backend

If an old broken volume is in a bad state you can reset DB data only with docker compose down and removing the postgres volume (this wipes panel data).

Sorry for the breakage — and thanks for the screenshot, it made the failure mode obvious.

[karutoil]

Update: fixed image is now on GHCR

CI is green and ghcr.io/catalystctl/catalyst-backend:latest has been republished (built 2026-08-10 from the packaging fix).

Verified on the published image:

• import('dotenv/config') resolves
• import('fastify') / @prisma/client resolve
• Prisma CLI present for migrate-on-boot
• relative ESM imports rewritten (./db.js)

Pull and restart:

cd catalyst-docker
docker compose pull
docker compose up -d
docker compose logs -f backend

You should see migrations apply and the backend stay healthy — no more Cannot find package 'dotenv'.

[karutoil]

@Dziondzio @HesusAIM

If you have any issues, please reach out on discord if possible!

https://discord.gg/mybxhmru3y