Cloudflare Error Codes: 10xx Blocks and 520-524 Server Errors

Every Cloudflare error falls into one of four groups: access denied, rate limiting, challenges, and origin and DNS errors. A proxy changes only your IP, network and country, so it helps only where the rule checks one of those.

Codes, titles and HTTP statuses are Cloudflare's (1xxx errors, 5xx errors); the "Proxy helps?" column is ProxyField's assessment, not Cloudflare's. The rule of thumb: repeated 403, 429 or 10xx pages mean the site is limiting you; 52x pages mean the site is broken. For 429 and Retry-After, see 429 Too Many Requests; for what a proxy cannot change in general, see what proxies can and cannot fix.

• The number. "Error 1020", "Error 522" and so on. It names the rule or the failure; Cloudflare documents each one under HTTP status codes.
• The Ray ID. A reference for that request, also sent as the cf-ray response header. You can't decode it, but the site owner can look it up in their Cloudflare account, so include it when you report a block.
• The diagram (on 52x pages). It shows "You", "Cloudflare" and "Host" (the website's server) and marks which link failed, usually "Host: Error".

The 10xx number sits in the HTML body; the HTTP status underneath is a normal one: 403 for most blocks, 429 for 1015, 409 or 530 for some DNS and setup errors, 5xx for server failures. Your scraper sees the status; the page body has the Cloudflare number.

Log four things per failed response: the status, the cf-mitigated header (Cloudflare sets it to challenge on every challenge page), the Retry-After header, and the error number in the body.

import re
import requests

PROXY = "http://<login>-country-us-city-new_york:<password>@gate.proxyfield.io:7000"
r = requests.get("https://example.com/", proxies={"http": PROXY, "https": PROXY}, timeout=30)

code = re.search(r"Error (1\d{3}|5\d{2})", r.text)
print(r.status_code,                       # 403, 429, 52x ...
      r.headers.get("cf-mitigated"),       # "challenge" on a challenge page
      r.headers.get("retry-after"),        # seconds to wait after a 429
      r.headers.get("cf-ray"),             # the Ray ID for the site owner
      code.group(1) if code else None)     # the Cloudflare number, e.g. 1020 or 522

The same check with cURL: curl -s -o /dev/null -D - https://example.com/ | grep -i -E '^(HTTP|cf-|retry-after)'.

The 10xx codes visitors and scrapers trigger, with Cloudflare's titles and HTTP statuses (1xxx errors):

Each of these, with step-by-step fixes, is in Cloudflare access denied errors. The other 10xx codes are configuration problems on the site's side: 1000 (DNS points to prohibited IP, 403), 1001 (DNS resolution error, 409), 1003 (direct IP access not allowed, 403), 1016 (origin DNS error, 530) and 1018 (could not find host, 409).

The site's, in every normal case; a scraper contributes only by overloading a heavy page with many threads. A 5xx page means your request reached Cloudflare and the failure is between Cloudflare and the website's server. If the site were blocking you on purpose, you would see 403, 429 or a 10xx number instead.

How to tell the cases apart in your logs:

One check settles most cases: load the same URL a few minutes later from a normal browser on another network. If it fails there too, it is the site.

Titles and causes from Cloudflare's per-code pages. None of these is caused by your proxy, your IP or your scraper's headers, so changing them does nothing. What you can do as a scraper:

1. Retry later, with back-off (for example 30 s, then 60 s, then 120 s; this pattern is ordinary practice, not a Cloudflare figure), and stop after a few tries.
2. Check whether the whole site is down from another network or a normal browser; if it is, wait.
3. Check Cloudflare's status page when many unrelated sites fail at once.

If you run the site, Cloudflare's 520 page gives these checks:

• As a temporary workaround, set the affected DNS record to DNS-only, or pause Cloudflare, and see whether the error follows the server.
• Turn off HTTP/2 to Origin under Speed > Settings > Protocol Optimization if the server's HTTP/2 setup is in doubt.
• In Origin Analytics, compare originResponseStatus with edgeResponseStatus: a 200 from the server served as a 520 means a malformed answer, such as oversized headers or an early close.
• In Logpush, OriginResponseStatus 0 is a failed connection only when CacheStatus is "miss" or "expired".
• For support, send the URL, the Ray ID, the output of /cdn-cgi/trace on your domain and two HAR files, one with Cloudflare on and one with it paused.

For a 403 with a Cloudflare page, start with Cloudflare 403 Forbidden. For why Cloudflare challenges a client in the first place, see why Cloudflare thinks you are a bot.

• A 10xx page: turn off your VPN or proxy, update your browser, pause extensions, and wait a few minutes after a 1015. If it persists, send the Ray ID to the site owner.
• A 52x page: the site is having problems. Reload in a few minutes; nothing on your device causes it.
• A "Just a moment..." page that loops: allow JavaScript and cookies for the site and try another browser.