generated: '2026-09-05' method: searched source: >- https://docs.cloudera.com/cdp-public-cloud/cloud/api/topics/mc-api-overview.html (the whole of Cloudera's API documentation), the 19 published CDP Swagger definitions, and live header inspection of https://api.us-west-1.cdp.cloudera.com on 2026-09-05 limit_count: 0 documented: false limits: [] response_headers: observed: [] checked: - X-RateLimit-Limit - X-RateLimit-Remaining - X-RateLimit-Reset - RateLimit - RateLimit-Policy - Retry-After note: >- None present. The full response header set observed on POST https://api.us-west-1.cdp.cloudera.com/api/v1/iam/listUsers was date, content-type, content-length, connection, x-cdp-request-id, content-security-policy, feature-policy, referrer-policy, x-content-type-options, x-frame-options, cache-control, pragma, x-request-id and Strict-Transport-Security. No rate-limit family is emitted. exhaustion_status: unknown exhaustion_note: >- No 429 is declared anywhere in the contract — in fact no 4xx status at all is declared on any of the 778 operations, which each list only 200 and `default`. Whether the control plane throttles, and what it returns when it does, is not discoverable from anything Cloudera publishes. finding: >- AN HONEST ZERO. Cloudera documents no rate limits for the CDP control plane API and emits no rate-limit signalling headers. This is a real gap for agent use: an agent driving 778 operations — many of which are polled repeatedly while provisioning completes — has no published budget, no runtime remaining-count, and no Retry-After to back off against. It cannot distinguish throttling from any other failure, because the error envelope is {code, message} with no documented code vocabulary either. adjacent_limits: note: >- Cloudera does document quotas and throttling extensively for components INSIDE a cluster — HBase throttle quotas, Kafka client quotas, YARN queue capacity — and the control plane itself exposes account-level service quotas through the compute and dw services. Those are resource quotas on what you may provision, not request-rate limits on the API, and they are not recorded as rate limits here. in_contract: >- "quota" appears 63 times across the harvested definitions — as request and response FIELDS (QuotaConfig, QuotaRequest) on the compute, de, dw, environments and ml services, never as an operationId. There is no operation for reading your API request budget. what_would_fix_it: >- Publishing a per-key request budget and emitting RateLimit / RateLimit-Policy (RFC 9331 style) or X-RateLimit-* headers plus Retry-After on 429. Cloudera already returns x-cdp-request-id on every response, so the header plumbing exists.