What causes a 500, how it differs from 502 and 503, and what to put in the body
Runs in your browser. Nothing you type is sent anywhere. Updated on 25 Oct 2026.A 500 is the code you return when something broke inside your own process and you have nothing more useful to say. It means an unhandled exception, not a caller mistake, so never send it for bad input. If a proxy in front of you could not reach your process at all, that is a 502; if you are up but refusing work, that is a 503. Put a log identifier in the body, never a stack trace.
Open the page with 500 Internal Server Error already selected.
Read "When to use it": an unhandled exception inside your own server, never a mistake by the caller.
Select 502 or 503 in the Codes list to check whether a proxy or an overloaded server is the real source.
Read the note: log an identifier and return it in the body instead of a stack trace.
This HTTP status codes reference lists every registered code from 100 to 511, plus the unregistered ones that nginx, Apache, Laravel and Cloudflare return. Each entry says what the code actually means, when to send it, whether a cache may store the response without cache headers, whether a redirect changes the request method, and which RFC defines it.
Two details most lists get wrong are included here. Cacheability is not a yes or no: RFC 9110 section 15.1 names twelve codes a cache may store with no explicit headers at all, which is why an accidental 404 can stick. And redirects differ: 301 and 302 let a client turn a POST into a GET, 303 always does, while 307 and 308 must repeat the method and body exactly.
Every common code comes with a copyable snippet for raw HTTP, Express, Fastify, FastAPI, Go and PHP, including the header that code requires, such as WWW-Authenticate on a 401 or Allow on a 405. Nothing is sent anywhere; the whole table is in the page.
The rest of the article, and all 6 questions about HTTP Status Codes, are on the HTTP Status Codes page.
Use 401 Unauthorized. It means the server does not know who you are, so a missing, expired or wrong credential belongs there. Use 403 Forbidden only when you know who the caller is and are still refusing. The specification wants a WWW-Authenticate header on every 401.
RFC 9110 section 15.1 lists 200, 203, 204, 206, 300, 301, 308, 404, 405, 410, 414 and 501 as heuristically cacheable, and RFC 7725 adds 451. A cache may store those with no Cache-Control or Expires header, which is why a temporary 404 can persist. Every other code needs explicit cache headers.
418 I'm a Teapot comes from RFC 2324, an April Fools' joke from 1998 about a coffee pot protocol. A 2017 proposal to reuse the number was withdrawn after objections, so it stays reserved in the IANA registry and must never be used for a real meaning.
No. 499 is an nginx log-only code for a client that hung up, and 520 to 527 are Cloudflare codes for problems between the edge and your origin. None are registered with IANA. They are listed here because you will meet them in logs, each marked with the product that invented it.
Still stuck? Ask about HTTP Status Codes and anybody who knows can answer. No account needed.
On the related tools: Hadolint Online – Run hadolint rules on a Dockerfile in your browser, with no install and no Docker · Readability Checker – Flesch reading ease and grade level, with what to change.
In the collection Debug and Inspect.