Hi, Thanks for the detailed report. I put together a minimal repro app and deployed it to test the same code path (raw HTTPS to googleapis.com, the OAuth token exchange via plain https, and the exact `admin.app().options.credential.getAccessToken()` call) — all three passed cleanly with no errors, sequentially and on a fresh deploy. That tells us the failure isn't a blanket Railway-to-Google connectivity issue, so it's likely tied to something specific about your app or deployment conditions. Could you share a few things so we can narrow it down? 1. **Dependency versions** — output of `npm ls firebase-admin google-auth-library gaxios` (or the relevant lines from your lockfile), this tells me about the versions of libraries you are using so that I can replicate this 1-1. 2. **Traffic pattern when it fails** — does it happen under concurrent/burst load, or even on a single isolated request? 3. **Idle time before failure** — does it tend to happen on the first request after the service has been idle for a while, or can it happen anytime? 4. **Frequency** — does every call fail now, or is it intermittent? 5. **Railway service details** — region, whether you're on a recent redeploy with a bumped Node/base image version, and roughly what time on July 3 it started 6. **Full error object** if you can share it (redact any token values) — specifically whether `err.cause` is populated, since that often points to the underlying socket error code 7. **Approximate request volume** your service handles — does this call happen once per request, or is it cached/reused across requests? Here's the minimal server I am using to replicate: https://github.com/mmjsmohit/fb-railway-repro

Hi, Thanks for investigating this. Here are the requested details: 1. Dependency versions firebase-admin@13.6.0 google-auth-library@9.15.1 gaxios@6.7.1 Full output: backend@1.0.0 +-- firebase-admin@13.6.0 | +-- @google-cloud/firestore@7.11.6 | | `-- google-auth-library@9.15.1 | +-- @google-cloud/storage@7.18.0 | | +-- gaxios@6.7.1 | | `-- google-auth-library@9.15.1 | `-- google-auth-library@9.15.1 | +-- gaxios@6.7.1 | +-- gcp-metadata@6.1.1 | `-- gtoken@7.1.0 | `-- gaxios@6.7.1 2. Traffic pattern The issue happens even on a single isolated request. No concurrent or burst traffic is required. 3. Idle time It is not related to idle time. It happens immediately after deployment as well. 4. Frequency It is 100% reproducible now. Every call to admin.app().options.credential.getAccessToken() fails. 5. Railway service details Region: Southeast Asia (Singapore) The issue started on July 3 around 10:00 PM IST. I don't remember intentionally changing the Node version or deployment configuration before it started. 6. Error The failure occurs inside admin.app().options.credential.getAccessToken(). The underlying error is: ❌ Firebase access token test failed: GaxiosError: Invalid response body while trying to fetch https://www.googleapis.com/oauth2/v4/token: Premature close at Gaxios._request (/app/node_modules/gaxios/build/src/gaxios.js:149:19) at process.processTicksAndRejections (node:internal/process/task_queues:103:5) at async GoogleToken._GoogleToken_requestToken (/app/node_modules/gtoken/build/src/index.js:241:19) at async GoogleToken._GoogleToken_getTokenAsync (/app/node_modules/gtoken/build/src/index.js:160:16) at async JWT.refreshTokenNoCache (/app/node_modules/google-auth-library/build/src/auth/jwtclient.js:173:23) at async JWT.refreshAccessTokenAsync (/app/node_modules/google-auth-library/build/src/auth/oauth2client.js:247:19) at async JWT.getAccessTokenAsync (/app/node_modules/google-auth-library/build/src/auth/oauth2client.js:276:23) at async ServiceAccountCredential.getAccessToken (/app/node_modules/firebase-admin/lib/app/credential-internal.js:137:9) at async /app/server/authMiddleWare.js:13:19 { config: { method: 'POST', url: 'https://www.googleapis.com/oauth2/v4/token', data: { headers: { 'Content-Type': 'application/x-www-form-urlencoded', 'User-Agent': 'google-api-nodejs-client/9.15.1', error: FetchError: Invalid response body while trying to fetch https://www.googleapis.com/oauth2/v4/token: Premature close 'x-goog-api-client': 'gl-node/22.23.0', Accept: 'application/json' }, responseType: 'json', retryConfig: { httpMethodsToRetry: [Array], currentRetryAttempt: 2, retry: 3, noResponseRetries: 2, retryDelayMultiplier: 2, timeOfFirstRequest: 1783106710973, totalTimeout: 9007199254740991, maxRetryDelay: 9007199254740991, statusCodesToRetry: [Array] }, paramsSerializer: [Function: paramsSerializer], body: '<<REDACTED> - See errorRedactor option in gaxios for configuration>.', validateStatus: [Function: validateStatus], errorRedactor: [Function: defaultErrorRedactor] }, response: undefined, at Gunzip.<anonymous> (/app/node_modules/node-fetch/lib/index.js:400:12) at Gunzip.emit (node:events:531:35) at emitErrorNT (node:internal/streams/destroy:170:8) at emitErrorCloseNT (node:internal/streams/destroy:129:3) at process.processTicksAndRejections (node:internal/process/task_queues:89:21) { type: 'system', errno: 'ERR_STREAM_PREMATURE_CLOSE', code: 'ERR_STREAM_PREMATURE_CLOSE' }, code: 'ERR_STREAM_PREMATURE_CLOSE', [Symbol(gaxios-gaxios-error)]: '6.7.1' } 7. Request volume This service is not under heavy load. Firebase Admin only requests an access token during authentication and related Firebase operations. Additional information: The exact same codebase, package-lock, Firebase service account JSON, and environment variables work perfectly on localhost. On localhost, admin.app().options.credential.getAccessToken() succeeds. On Railway, the same call consistently fails with ERR_STREAM_PREMATURE_CLOSE. One additional observation: I found another Railway Station thread where another developer reported an almost identical error (ERR_STREAM_PREMATURE_CLOSE while requesting https://oauth2.googleapis.com/token) and resolved it by replacing the internal node-fetch token request with Node's global fetch (undici). My symptoms appear to match that report very closely. Please let me know if you'd like me to test a specific runtime, dependency version, or any additional diagnostics.

Update node version to `24.18` or newer

Add a variable `RAILPACK_NODE_VERSION=24.18` to update node version on railpack