zulip/templates/zerver/api/rest-error-handling.md

2.3 KiB

Error handling

Zulip's API will always return a JSON format response. The HTTP status code indicates whether the request was successful (200 = success, 40x = user error, 50x = server error). Every response will contain at least two keys: msg (a human-readable error message) and result, which will be either error or success (this is redundant with the HTTP status code, but is convenient when printing responses while debugging).

For some common errors, Zulip provides a code attribute. Where present, clients should check code, rather than msg, when looking for specific error conditions, since the msg strings are internationalized (e.g. the server will send the error message translated into French if the user has a French locale).

Each endpoint documents its own unique errors; below, we document errors common to many endpoints:

{generate_code_example|/rest-error-handling:post|fixture(400)}

{generate_code_example|/rest-error-handling:post|fixture(401)}

{generate_code_example|/rest-error-handling:post|fixture(403)}

{generate_code_example|/rest-error-handling:post|fixture(429)}

The retry-after paremeter in the response indicates how many seconds the client must wait before making additional requests.

To help clients avoid exceeding rate limits, Zulip sets the following HTTP headers in all API responses:

  • X-RateLimit-Remaining: The number of additional requests of this type that the client can send before exceeding its limit.
  • X-RateLimit-Limit: The limit that would be applicable to a client that had not made any recent requests of this type. This is useful for designing a client's burst behavior so as to avoid ever reaching a rate limit.
  • X-RateLimit-Reset: The time at which the client will no longer have any rate limits applied to it (and thus could do a burst of X-RateLimit-Limit requests).

Zulip's rate limiting rules are configurable, and can vary by server and over time. The default configuration currently limits:

  • Every user is limited to 200 total API requests per minute.
  • Separate, much lower limits for authentication/login attempts.

When the Zulip server has configured multiple rate limits that apply to a given request, the values returned will be for the strictest limit.

Changes: The code field in the response is new in Zulip 4.0 (feature level 36).