Hey! What are you trying to deploy and can you share the full build output before failure?

Did you recently change anything before this issue? Like build commands, dependencies, railway.json...?

Hi, thanks for the quick response!

What I'm deploying: A Next.js 14.2.35 app (App Router, TypeScript), deployed via the CLI (railway up), using the Dockerfile builder (auto-detected, no railway.json/railway.toml). Multi-stage Dockerfile (builder + runner), Node 20-bullseye-slim base image. Project: dece-sistema, project ID 5a7aec9e-2335-46eb-97aa-ed575fd9bf92, service ID 36d685d2-968e-4de4-9b6a-05ef63a71597.

Recent changes: No changes to build commands, dependencies, or config files caused this — it started when deploying a routine code update (new source files added to an existing, previously-working app). I did later try raising NODE_OPTIONS max-old-space-size from 2048 to 4096 in the Dockerfile, and adding experimental.cpus: 1 / workerThreads: false to next.config.js, as troubleshooting attempts — neither changed the failure pattern at all.

The problem: Builds fail with "Deploy failed" and no application-level error message at all (no TypeScript error, no webpack error, nothing) — I've confirmed locally that npm run build completes successfully every time with the exact same code. The failure point is inconsistent across attempts — sometimes right after "unpacking archive", sometimes during a Docker COPY layer, sometimes during npm run build itself at "Creating an optimized production build...". It's happened on ~12 consecutive attempts since Aug 25, almost always scheduled on "builder-vdsuic" (one attempt also failed on "builder-eaxqtv"). This timing matches the "Delayed deploys in AMS" (Aug 25) and "Deployments are slow" (Aug 26) incidents on your own status page.

Recent failed deployment IDs for reference: ed757782-e246-4305-abea-a8aa5f443278, d6da89ac-89ff-4cf8-8737-8a12f2dfbac6, 14ffb3a8-8364-4d41-845f-50d1a0900e95, bf01197e-b755-4c3f-8bcd-d1cb840ef5ac

Could you check what's going on with builder-vdsuic for this project? Happy to share more deployment IDs if useful. Thanks!

Following up with additional diagnostic evidence from further testing:

I deliberately lowered NODE_OPTIONS=--max-old-space-size to an artificially low 512MB to force Node's own graceful OOM error ("JavaScript heap out of memory") to surface if this were a real V8 heap memory issue. The build still failed silently at the same point ("Creating an optimized production build...") with no error message at all — Node's own OOM error never appeared, even at this very low limit. This strongly indicates the build process is being killed externally (not failing due to Node/V8 running out of its own heap).

I also tried limiting Next.js's build-time parallelism (experimental.cpus: 1, workerThreads: false) and separately setting RAYON_NUM_THREADS=1 to constrain the native SWC compiler's internal thread pool (which is not governed by NODE_OPTIONS). Neither change had any effect — the build still fails at the exact same step.

I disabled the build cache entirely (NO_CACHE=1) to rule out a corrupted cache layer. The build ran fully fresh (confirmed by full apt/npm reinstall in the logs) and still failed identically.

Given that memory limits, build caching, and both levels of build parallelism (Next.js and native SWC) have all been ruled out, this appears to be an infrastructure-level issue with the build environment itself, not something fixable from the application or Dockerfile side. Could someone take a look at the specific builder machine/environment assigned to my recent deploy attempts? Happy to provide more deployment IDs if useful.