18 days ago
My Dockerfile deploy fails at the final build step (a Medusa v2 / Vite admin build) with NO error in the build logs. The build log shows [6/8] RUN npm ci completing fine, then [8/8] RUN npm run build starts and the log just stops — no error message, no exit code, no "did not complete" line, no heap-out-of-memory. The same Dockerfile builds successfully on my local machine (docker build, exit 0). Service is in US West.
Can you tell me WHY the builder is killing this build? Specifically: is it an out-of-memory kill, a build timeout, or a plan/builder limit? The plan lists 48GB RAM per service but I can't tell what the build step itself is allocated. Deployment ID for the latest failure: b565c561 (project axe-store).
1 Replies
Status changed to Awaiting Railway Response Railway • 18 days ago
17 days ago
This thread has been opened as a bounty so the community can help solve it.
Status changed to Open Railway • 17 days ago
15 days ago
Your build isn't crashing — it's getting killed for using too much memory, and the silent log is the giveaway.
When a build runs out of memory, the system force-kills it instantly (SIGKILL). That kind of kill can't be caught, so Node/Vite never gets a chance to print an error. The log just... stops. That's exactly what you're seeing.
The 48GB on your plan is a trap here — that's the memory for your running app, not for the build. The builder runs on totally separate, smaller hardware (often just 4–8GB). So you can have a huge plan and still get killed mid-build. Medusa v2's admin build (Vite/Rollup) is famously memory-hungry, which is why it builds fine on your local machine (more RAM) but dies on the builder.
It's probably not a timeout — those usually leave a "build exceeded time limit" message