Hi Railway team,

Since about 10:50 UTC today (2026-11-05), every build in my project fails

before any build step runs, with:

Build Failed: build daemon returned an error < failed to solve:

frontend grpc server closed unexpectedly >

Services: journal and sales-agent

Builder: Railpack 0.43.0 (railpack.json in repo), GitHub source

Failed deployments (all FAILED during BUILDING, triggered by commit 6a8560e7):

journal 6b404e95-2f8b-44bc-8165-e71c3c6180fc 10:52 UTC

sales-agent 6c7f8791-d85b-4384-bf66-f12681589829 10:52 UTC

journal 063e7f31-486e-4ee1-8d56-f68ca7e41af6 10:59 UTC (retry)

sales-agent 577d5717-f830-466e-98ab-28da5e24e8dd 10:59 UTC (retry)

Last good build: journal ebfab7fe-898f-4f41-9962-d16cbbde2322, SUCCESS at

08:19 UTC today, built from the previous commit 5d02cf03.

What the logs show: the Railpack plan prints normally (python 3.13.17,

uv 0.13.5, install: uv sync --locked ...), "resolve image config for

railpack-frontend:v0.43.0" runs, and then the daemon crashes. In the

successful build the next line is "load build definition from

./railpack-plan.json"; in the failed ones that line never appears and the

build ends within about 40-100 s.

What I have ruled out:

• Same frontend image digest in the good and failed builds

  (railpack-frontend:v0.43.0@sha256:355ef61fae6f2551103cd8b10c710cb07167dd353d561f15c69b8d7ac1c676cc).

• No change to railpack.json, railway.json, pyproject.toml, uv.lock or

  .dockerignore between the good commit and the failing one.

• Both services fail identically within the same second, and a retry about

  7 minutes later fails the same way.

• Production still runs the last good deployment, so no outage on our side.

Questions:

1. Is there a problem with the build daemon / Railpack frontend on the

   builder these deployments landed on? Can you look at the four

   deployments above?

2. Is this related to the incident on status.railway.com (Nov 4, 22:03 UTC)?

3. Is there anything I can change on my side (for example a different

   builder) that you would recommend while this is investigated?

Thanks,

Asher