Skip to content
Quiltpro pre-deploy networking failure after subscription reactivation
kelliottmagic
OP

20 days ago

This appears similar to the current report “Pre-deploy command cannot reach private Postgres”. Please inspect the pre-deploy container’s private-network access and clear the pending zero-change commitChanges workflow. Ordinary redeploys continue to reuse the same failing state.

Solved$20 Bounty

10 Replies

Railway
BOT

20 days ago

Your recent Quiltpro deployments are not getting as far as networking: each one fails while the container is being started, with the error that the executable cd was not found. cd is a shell builtin rather than a program, so a start or pre-deploy command that begins with cd ... cannot be run directly and fails before any connection to Postgres is attempted, which is why every redeploy hits the same error.

The commands set in the service's dashboard settings don't contain cd (the start command is empty and the pre-deploy command is npx prisma migrate deploy), so the command starting with cd is coming from somewhere else, most likely a config file in your repository. Either remove the cd or wrap the whole command in a shell, for example sh -c "cd app && npx prisma migrate deploy".

As for the zero-change changeset, it is staged edits that were never applied, not a workflow that is running. It has no effect on deployments and doesn't need to be cleared.


Status changed to Awaiting User Response Railway • 20 days ago


Railway

Your recent Quiltpro deployments are not getting as far as networking: each one fails while the container is being started, with the error that the executable `cd` was not found. `cd` is a shell builtin rather than a program, so a start or pre-deploy command that begins with `cd ...` cannot be run directly and fails before any connection to Postgres is attempted, which is why every redeploy hits the same error. The commands set in the service's dashboard settings don't contain `cd` (the start command is empty and the pre-deploy command is `npx prisma migrate deploy`), so the command starting with `cd` is coming from somewhere else, most likely a config file in your repository. Either remove the `cd` or wrap the whole command in a shell, for example `sh -c "cd app && npx prisma migrate deploy"`. As for the zero-change changeset, it is staged edits that were never applied, not a workflow that is running. It has no effect on deployments and doesn't need to be cleared.

kelliottmagic
OP

20 days ago

Thank you. I checked the repository at the deployed main branch. There is no Railway configuration, Procfile, Nixpacks file, package script, Docker command or relevant Git history containing a deployment command beginning with cd. The Dockerfile uses WORKDIR /app/server and CMD ["node", "server.js"].

The dashboard pre-deploy command is npx prisma migrate deploy, and the dashboard start command is empty. Please inspect deployment 64e496be and identify the exact service manifest field containing the cd command and where Railway sourced it. It appears to be stale Railway configuration rather than repository configuration.


Railway
BOT

20 days ago

This thread has been marked as private.

Status changed to Awaiting Railway Response Railway • 20 days ago


Railway
BOT

20 days ago

That deployment built successfully and then stopped in the pre-deploy step about seven seconds in, because the command it was given starts with cd, which cannot be run on its own, so it never reached the point of connecting to Postgres.

Our earlier reply pointed to a config file in your repository, but your check found none, and the service's current settings show no cd either: the pre-deploy command is npx prisma migrate deploy, the start command is empty, and the builder is set to Railpack rather than Dockerfile. We have not been able to establish from this deployment's record where the cd command was sourced from.


Status changed to Awaiting User Response Railway • 20 days ago


kelliottmagic
OP

20 days ago

Additional evidence: GitHub shows commit 5c80fe2 deployed successfully to this exact Railway production environment on 5 July. I compared that commit with current main (96036cc). The Dockerfile, .dockerignore, server/package.json, lockfile and deployment configuration are unchanged. The only subsequent changes are in four client UI files.

Therefore, the failing command beginning with cd is not present in the repository and was not introduced by the current commit. Please inspect the failed deployment’s baked serviceManifest and identify the Railway-side configuration field supplying that command.


Status changed to Awaiting Railway Response Railway • 20 days ago


Railway
BOT

20 days ago

One detail matters for your comparison: this service is set to build with Railpack, not your Dockerfile, so the Dockerfile's WORKDIR and CMD ["node", "server.js"] are not what this deployment runs. Unchanged Dockerfile and repository files therefore don't rule out the start command that ends up in the image.

The failed deployment's record shows it stopping when the container is started, with the error that cd cannot be found as an executable. The service's saved settings still show an empty start command and npx prisma migrate deploy as the pre-deploy command, so neither saved field contains cd. The deployment record does not show which field supplied the cd command, and we have not established its source.


Status changed to Awaiting User Response Railway • 20 days ago


Railway

One detail matters for your comparison: this service is set to build with Railpack, not your Dockerfile, so the Dockerfile's `WORKDIR` and `CMD ["node", "server.js"]` are not what this deployment runs. Unchanged Dockerfile and repository files therefore don't rule out the start command that ends up in the image. The failed deployment's record shows it stopping when the container is started, with the error that `cd` cannot be found as an executable. The service's saved settings still show an empty start command and `npx prisma migrate deploy` as the pre-deploy command, so neither saved field contains `cd`. The deployment record does not show which field supplied the `cd` command, and we have not established its source.

kelliottmagic
OP

20 days ago

The affected deployment’s Build Logs explicitly show internal load build definition from Dockerfile, so the Dockerfile was used despite the saved builder being labelled Railpack. Railway’s documentation also says a root file named Dockerfile is detected automatically.

The automated replies have now confirmed that neither the repository, saved start command nor saved pre-deploy command contains cd, but the deployment was nevertheless given a command beginning with cd. Please escalate this to a Railway team member who can inspect the deployment’s complete generated service manifest and clear or correct the stale command. I do not want to alter the database or move migrations into the application start command.


Status changed to Awaiting Railway Response Railway • 20 days ago


Railway
BOT

20 days ago

This is the kind of problem the Railway community is well placed to help with, so we'd like to open your thread as a community bounty. Railway pays a bounty to the community member who solves it, and threads like this usually get picked up quickly.

Opening it makes this entire thread public, including everything already posted. Nothing becomes public until you decide. Use the buttons below.

  • Open to the community - Before you click, take a moment to edit or remove anything you'd rather not share. The thread becomes publicly visible right away.
  • Keep it private and close the thread - Nothing becomes public and the thread closes.

Status changed to Awaiting User Response Railway • 20 days ago


Railway
BOT

20 days ago

This thread has been opened as a public bounty so the community can help solve it. The thread and any further activity are now visible to everyone.

Status changed to Open Railway • 20 days ago


Just to clarify, your pre-deploy command isn't able to reach Postgres over the private network? (With both services being in the same project and environment)


0x5b62656e5d

Just to clarify, your pre-deploy command isn't able to reach Postgres over the private network? (With both services being in the same project and environment)

kelliottmagic
OP

20 days ago

Both Quiltpro and Postgres are in the same project and production environment, and Postgres is healthy. We initially suspected private-network access, but Railway’s deployment inspection reported that deployment 64e496be fails before attempting a Postgres connection because the command it receives begins with cd, which is being treated as an executable.

The saved pre-deploy command is only npx prisma migrate deploy, the saved start command is empty, and there is no deployment command beginning with cd anywhere in the repository. The unresolved question is where that generated cd command came from in the deployment’s service manifest. If that is corrected and Prisma then cannot reach Postgres, we can investigate private networking separately.


kelliottmagic
OP

20 days ago

Resolved by creating a fresh Railway service linked to the same GitHub repository and existing Postgres database. The new service successfully ran npx prisma migrate deploy and deployed normally, confirming the failure was specific to the old service configuration. Thank you for the help.


Status changed to Solved kelliottmagic • 20 days ago


Welcome!

Sign in to your Railway account to join the conversation.

Loading...