Intermittent Vercel server connection errors causing Googlebot crawler failures

Help
, , ,
1

We are seeing intermittent crawler failures where Google cannot fetch our robots or pages (Search Console shows “Server connection error” / “Not fetched”), causing pages to be reported as “Not available to Google”.

Site

Problem summary

  • Google Search Console URL Inspection and robots checks report intermittent failures (robots.txt: alternating “Fetched” / “Not Fetched”; /services live test: “Failed: Server connection error”).

  • Failures are intermittent (sometimes OK, sometimes Not Fetched). The files are static in public and middleware is excluded from static routes, so this looks like an edge/network or POP-level issue.

Evidence and observed successful fetches (please inspect nearby logs for failures)

  • Manual fetches performed now (Fri, 10 Apr 2026 17:05:38 GMT):

    • GET /health.txt — 200 OK
      X-Vercel-Id: yul1::cj29j-1775840737891-5cd7918a8377
      Date: Fri, 10 Apr 2026 17:05:38 GMT

    • GET /robots.txt — 200 OK
      X-Vercel-Id: yul1::bcspj-1775840738786-661d683fe6c3
      Date: Fri, 10 Apr 2026 17:05:38 GMT

    • GET /services — 200 OK
      X-Vercel-Id: yul1::iad1::bcspj-1775840738922-958180555753
      Date: Fri, 10 Apr 2026 17:05:39 GMT

Search Console evidence

  • Live test failure shown in Search Console: Tested on Apr 10, 2026, 12:52 PM — “Page fetch: Failed: Server connection error”.

  • Robots.txt history shows multiple “Not Fetched” entries across Apr 9–10, 2026 (see attached screenshots).

Related community thread

  • See this Vercel community discussion which describes the same symptom when robots.txt is served by an Edge Function that assumes certain headers (e.g., accept-language) are present and fails when they are missing: https://community.vercel.com/t/unreachable-robots-txt/386
    Please also check for EDGE_FUNCTION_INVOCATION_FAILED or missing-header handling in the edge logs (Google crawlers sometimes omit accept-language).

Requested actions

  1. Search Vercel edge logs for requests to this domain from Google user-agents (Google Inspection Tool / Googlebot) for Apr 9–10, 2026 and identify any transient errors, connection timeouts, or network failures.

  2. Inspect logs specifically around the X-Vercel-Id values above and any nearby failing requests; return any error messages or reasons the edge returned a connection error to Google (network errors, origin timeouts, edge instance errors, rate-limiting, SSL issues, etc.).

  3. Confirm whether this was an edge/network transient issue and if there is an ongoing incident. If a specific POP or node is problematic, advise mitigation steps.

  4. If you need additional X-Vercel-Id values or exact UTC timestamps from Search Console failures I will provide them.

Attachments I’m including

  • Search Console screenshots showing the live test failure and robots.txt versions.

  • The X-Vercel-Id values captured above.

Thanks — please let me know what you find or if you need more precise timestamps or additional request IDs.