Developer checklist

What to confirm before shipping an e6data integration - auth, base paths, polling, pagination, errors, idempotency, propagation, and secrets.

Before you ship an integration against the e6data API, confirm the following.

• Auth - choose the right credential (a service token) and send it as Authorization: Bearer . A bearer token beats a session cookie. See Authentication.
• Poll, don't assume - cluster operations and query execution are asynchronous; poll status to a terminal state.
• Errors - branch on the error code, not just the HTTP status.
• HTTP/2 - required only for the query path; management REST is plain HTTPS. Don't force curl's --http2 flag for management calls.
• Idempotency - repeating suspend, resume, and scale is safe; token creation is not (each call mints a new secret).
• Propagation - token revocations and permission changes are cache-backed and can take a short time to take effect; design retries and rotations around eventual, not instant, consistency.
• Secrets - store tokens in a secret manager, one per consumer, rotate on expiry, and never log or commit them.

See also

• Automate clusters, catalogs, schedules, and policies - the endpoints and recipes.
• Connecting to the engine - the query path details (HTTP/2, chunks, sessions).