Build failing repeatedly with truncated logs — inboxtool service
mandrycabriotenerife-collab
HOBBYOP

a month ago

Description: Service: inboxtool Project: peaceful-charm Environment: production Deployment ID: d352ff66-186d-4b92-9e02-2e13c6fe3256

Problem: Build fails 6+ times in a row on the same commit (8d00ed42). Build logs are truncated and contain only one line: "scheduling build on Metal builder". The actual error is not captured or displayed.

Code is correct: Commit 8d00ed42 explicitly fixes a syntax error in French translations (stray trailing commas in pick()/tr() calls) that was breaking previous builds. I verified the git diff — the fixes are correct.

What I've tried:

Redeployed 3+ times — still fails

Empty commit to force rebuild — still fails

Verified code syntax — all correct

Dockerfile.railway: npm ci && npm run build — unchanged, correct.

This appears to be a Metal builder infrastructure issue on Railway's side, not a code problem.

Awaiting Conductor Response$10 Bounty

4 Replies

Railway
BOT

a month ago

This thread has been opened as a bounty so the community can help solve it.

Status changed to Open Railway • about 1 month ago


mandrycabriotenerife-collab
HOBBYOP

a month ago

Please can a Railway staff member manually restart/reset the builder node for my service? It's been stuck on "scheduling build" for over a day. Deployment ID: d352ff66-186d-4b92-9e02-2e13c6fe3256, service: inboxtool, project: peaceful-charm.


mandrycabriotenerife-collab
HOBBYOP

a month ago

Checked .dockerignore — node_modules, .git, dist already excluded, so that's not it.

The project builds from a custom Dockerfile.railway (not Nixpacks/Heroku/Paketo), so I'd rather not switch the builder engine — that would likely break the build in a different way since it needs the multi-stage Docker/nginx setup.

Don't have Railway CLI set up, so couldn't try the --clear-cache option yet.

Redeployed 10+ times over the last day, including with a fresh empty commit and a region change — always dies instantly at "scheduling build on Metal builder" with zero compiler output.


mandrycabriotenerife-collab
HOBBYOP

a month ago

Update: Resolved.

Turns out there were two separate issues stacked on top of each other:

  1. A real syntax bug in the code (this was the actual fix needed). Several commits added a French translation as a 4th argument to pick(ru, en, es) / tr(ru, en, es) calls. In multi-line calls that already ended with a trailing comma before the closing ) (valid JS), the insertion script placed the new French argument after that comma instead of replacing it, producing an empty elided argument:

pick(

"ru text",

"en text",

"es text",

, "fr text" // ← stray comma before the new arg = syntax error

)

This is invalid JS and is exactly what showed up as [builtin:vite-transform] Error: Unexpected token. One related case also had a literal line break inside a French string instead of the

placeholder the other languages used, causing an "Unterminated string" error. Found and fixed every occurrence of this pattern across the whole codebase (14 files total).

  1. Separately, the Metal builder itself was stuck — for a while, builds weren't even starting (dying instantly at "scheduling build" with zero compiler output, before reaching any Dockerfile step), regardless of whether the code was valid. That appears to have cleared up on Railway's side.

Once both were addressed, the build went through cleanly. Thanks for the help investigating!


mandrycabriotenerife-collab
HOBBYOP

a month ago

It still open?


Welcome!

Sign in to your Railway account to join the conversation.

Loading...