Summary

website/docs/components/data-connectors/https/deployment.md (introduced in #1636) listed only 5 of the 14 rate-control metrics that the HTTP connector actually emits. Operators relying on the deployment guide to set up dashboards would miss the in-flight, permit, wait-duration, and Retry-After cooldown metrics — which are precisely the ones needed to diagnose rate-limit issues.

Changes

Added a Type column and 9 missing metric rows in the Metrics section:

• inflight_operations
• rate_control_available_permits
• rate_control_acquisitions_total
• rate_control_acquire_errors_total
• rate_control_wait_duration_ms
• rate_limit_retry_after_updates_total
• rate_limit_retry_after_waits_total
• rate_limit_retry_after_wait_duration_ms
• rate_limit_retry_after_remaining_ms

Names and descriptions copied verbatim from HTTP_RATE_CONTROL_METRIC_SPECS in source.

Reference

Verified against spiceai/spiceai trunk — crates/runtime/src/dataconnector/http_rate_control.rs lines 311-392 (HTTP_RATE_CONTROL_METRIC_SPECS array).

vNext-only — the https/deployment.md file does not exist in any website/versioned_docs/version-*/ directory.

Test plan

• cd website && npm run build passes locally
• All metric names match the literal strings in MetricSpec::new(...) calls
• No metric name was invented from context — every row corresponds to a verbatim source string