v0.30.6 image ships empty /app/node_modules/postgres/package.json → ERR_INVALID_PACKAGE_CONFIG

To Reproduce

1. Have a Dokploy instance running dokploy/dokploy:v0.30.6.
2. Force the container to be recreated — e.g. reboot the host, or run
   docker service update --force dokploy. An already-running container keeps
   working, so the bug only surfaces on recreation.
3. The dokploy service never reaches 1/1 and Swarm restarts it in a loop:

$ docker service ls
ID NAME MODE REPLICAS IMAGE
zyxnn22xn794 dokploy replicated 0/1 dokploy/dokploy:v0.30.6

4. Logs show Node failing before startup:
   [wait-for-postgres] Postgres is reachable
   Error: Invalid package config /app/node_modules/postgres/package.json while importing "postgres" from /app/dist/migration.mjs
   at Object.read (node:internal/modules/package_json_reader:110:33)
   at packageResolve (node:internal/modules/esm/resolve:769:43)
   at moduleResolve (node:internal/modules/esm/resolve:853:18)
   code: 'ERR_INVALID_PACKAGE_CONFIG'
   Node.js v24.4.0

5. The file is empty in the published image. This reproduces on a clean pull on any machine, so it is not local layer corruption

$ docker run --rm --entrypoint sh dokploy/dokploy:v0.30.6 -c 'wc -c /app/node_modules/postgres/package.json'
0 /app/node_modules/postgres/package.json

$ docker run --rm --entrypoint sh dokploy/dokploy:v0.30.5 -c 'wc -c /app/node_modules/postgres/package.json'
1814 /app/node_modules/postgres/package.json

Affected digest: sha256:1d6bd69ba58c1b4e305a9a33d77d8c3e0ee34707169f680cc600ab7ef1c3e6d8

Workaround: docker service update --image dokploy/dokploy:v0.30.5 dokploy

Current vs. Expected behavior

Expected: after the container is recreated, the dokploy service starts normally and reaches 1/1, exactly as it does on v0.30.5.

Current: the container exits with code 1 on every start, before the server ever comes up. Swarm keeps rescheduling it, so the service stays at 0/1 indefinitely and the dashboard is unreachable (curl http://localhost:3000 refuses the connection). Node fails while resolving the postgres import because the package's package.json in the image is 0 bytes.

The instance had been running fine for weeks. It only broke after an unattended kernel upgrade rebooted the host, which forced Swarm to recreate the container from the image. Postgres, Redis and Traefik all stayed healthy — only the dokploy service was affected.

Provide environment information

Operating System:
  OS: Ubuntu 24.04
  Arch: x86_64
  Kernel: 6.8.0-139-generic
Dokploy version: v0.30.6
Docker: Swarm mode, single node
Postgres: 16
VPS Provider: Hostinger
What applications/services are you tying to deploy?
  Two static frontend apps (unaffected - they kept running at 1/1).
  The failure is in the dokploy service itself, which never starts.

Which area(s) are affected? (Select all that apply)

Docker Compose, Docker

Are you deploying the applications where Dokploy is installed or on a remote server?

Remote server

Additional context

No response

Will you send a PR to fix it?

No

Start by comparing /app/node_modules/postgres/package.json in the v0.30.6 and v0.30.5 images, then inspect the image build inputs that produce /app/dist/migration.mjs and the dependency tree. Reproduce with the documented docker run and service recreation commands. Done means the published image contains a valid postgres package.json and the dokploy service starts at 1/1 without ERR_INVALID_PACKAGE_CONFIG.