Rate limits now return 429, and are counted per key

When a key goes over its rate limit, the API now answers 429 Too Many Requests with a Retry-After header, instead of 503 Service Unavailable:

HTTP/1.1 429 Too Many Requests
Retry-After: 60

{ "error": "Rate limit exceeded: 300 requests a minute for this key" }

Wait for the number of seconds in Retry-After before trying again. Previously an over-limit call got a 503, which looked like an outage. If your integration retries on 503, make sure it also retries on 429 after the Retry-After delay.

Limits are now counted per API key. Before, V2 calls were counted by IP address, so several keys calling from one server shared a single limit. Now each key has its own. The standard limit is 300 requests a minute. If you need more, contact us.

Calls with no key are still limited by IP address.