issue while deploying to gcp vm a turbo repo backend dockerized using pnpm

[kishu279]

Summary

turbo repo pnpm for the mono repo and packages contains the database, auth, shared, redis this contains the shared configurations or client like in prisma on database while dockerzing i am getting indefinite number of issues
need help
how does the production wise code base to be setted up

what i am doing according to the google where i found something and then added
building using tsup with config
`
import { defineConfig } from "tsup";

export default defineConfig({
entry: ["src/server.ts"],
format: ["cjs"],
platform: "node",
bundle: true,
noExternal: [/^@repo//],
external: ["@prisma/client/runtime/client"], // Prisma has .node binary — cannot bundle
shims: true, // ← injects createRequire so CJS dynamic requires work in ESM
clean: true,
outDir: "dist",
});

souravdeepakpoddar@shree
`
so that making inline into the dist/server.js the final outcome
except the client
and now i am getting another issue client runtime not present and then build failed
so i can add ths prisma/client/runtime execution on tsup will make this work

but getting different types of issues
can anyone tell me or can share the reference of docs so that i can make this done possibly make this work

what just i want is to make run the backend on the vm for now that to be prefered for the very first time then go to the production wize

Additional information

No response

Example

No response

[NspxMiguel]

The @prisma/client/runtime/client error is usually a packaging problem: the server bundle references Prisma, but the final image does not contain the generated client/runtime and its engine files. Do not add Prisma's internal runtime path to tsup to fix that; it is not a public entry point.

A reliable baseline for a pnpm Turborepo image is:

1. Run turbo prune <backend-workspace-name> --docker in a builder stage.
2. Copy out/json plus the pruned lockfile and run pnpm install --frozen-lockfile.
3. Copy out/full.
4. Run prisma generate from the database workspace inside the Linux image, then build the backend.
5. In the runtime stage, copy the built server, the generated Prisma client, and the production node_modules needed by that workspace.

Turborepo's Docker guide explains the split out/json / out/full flow: https://turborepo.com/docs/guides/tools/docker

Prisma also has a pnpm-workspaces guide showing a separate database package and an explicit generate step: https://docs.prisma.io/docs/guides/deployment/pnpm-workspaces

For the current Prisma generator, prefer a custom output directory in schema.prisma and import from that generated directory. If you use the older prisma-client-js generator, keep @prisma/client external to tsup and make sure it is installed in the runtime image. In either case, prisma generate must run after the schema has been copied.

Your tsup configuration alone is not enough to identify the remaining failure. If this still fails, please paste:

• the exact build/runtime error,
• the generator client block from schema.prisma,
• the backend and database workspace names from their package.json files, and
• the Dockerfile.

That will show whether the missing piece is the generated output, pnpm workspace deployment, or the platform-specific Prisma engine.