Summary

Every deployment on two services in the same project is stuck in INITIALIZING indefinitely with deploymentStopped: true in the deployment meta, zero configErrors, and no build ever associated. Reproduces on BOTH GitHub-triggered deploys AND railway up from CLI. Recreating services from scratch, reconnecting GitHub, and correcting root directory all did not fix it.

Deployment meta (via GraphQL)

status: "INITIALIZING" deploymentStopped: true canRedeploy: false meta.configErrors: null meta.reason: "deploy" meta.builder: "RAILPACK"

Build log says "Waiting for build to start…" forever. The deployment card shows "Deployment in progress: Taking a snapshot of the code…" indefinitely.

railway logs --build returns: Deployment does not have an associated build

What I've tried (all reproduce the same stuck state)

1. Original break (~2 days ago): deployments stopped auto-firing; Source panel showed Bad credentials. Tried service disconnect/reconnect, full GitHub App uninstall/reinstall, Railway OAuth revoke/re-auth — no change.
2. Deleted and recreated both services via Terraform with fresh service IDs. Source reconnected cleanly (the "Bad credentials" error is gone).
3. Root directory fix: corrected root_directory from /server and /agent (leading slash caused configErrors: ["Could not find root directory: /server"] on one FAILED deploy) to server and agent. New deploys have configErrors: null but still stuck INITIALIZING.
4. Bypassed GitHub entirely with railway up: deployment went to FAILED with the same deploymentStopped: true pattern, no build logs.
5. Removed stuck deployments and pushed fresh commits: webhooks fire, new deployment entries created with correct commit hashes within seconds — they immediately enter the same stuck INITIALIZING state. CI on GitHub Actions passes green.

What works vs what doesn't

• Other services in the same project (Postgres, Redis) healthy, SUCCESS
• GitHub webhook reaches Railway (deployment entries get created fast)
• Terraform / API calls to mutate service config succeed
• No deployment ever transitions past INITIALIZING on either app service, in either environment (staging or production)
• railway up from CLI produces same stuck/failed state

Earlier observed symptom

Dashboard briefly showed an Apply N changes / Deploy top banner with a red "An unknown error occurred" subtitle. Clicking Deploy did not clear it nor unstick deployments.

Is anyone else seeing this today?

This symptom pattern matches past incidents ("Railway Metal build slowdown", "Repository snapshot operation timed out"), but the public status page currently shows fully operational. Wondering if anyone else is currently hitting deploymentStopped: true without any visible error.

Happy to share project/service/deployment IDs and logs privately with a Railway staff member if this is a platform-side issue that needs investigation.