Get auto-tune status for a SQL warehouse

User-configured backoff thresholds for the 7-day post-apply monitor.

Mirrors Snowflake's BackoffConfiguration shape (percentages, not raw ratios) but adds bounds validation Snowflake's never had — a threshold below 110% would trigger backoff on ordinary noise, and an absurdly high one (past 1000%) defeats the monitor's purpose entirely. Both default to 150 (allow up to 1.5x the baseline before treating it as a regression).

The current actionable recommendation, if any.

The 7-day post-apply monitoring window, if one is in progress or has produced a stabilized/backed-off value.

The most recent audit event for this warehouse, for a lightweight "last changed" display without a full audit-history fetch.

Whether the SP can currently enable auto-tune on this warehouse.

Surfaced on the status response so the FE can render the Auto Tune Switch disabled with a persistent reason (instead of a momentary toast on a failed enable) when the service principal lacks CAN_MANAGE. Fail-open: any credential/read/timeout error leaves blocked=False (the hard enable-gate in the toggle path still protects the mutation).

Used in two positions. Top-level it is the AUTO STOP axis's own enable gate: CAN_MANAGE is folded in, plus Auto Stop's no_active_recommendation. It does not speak for Warehouse Sizing. can_manage_blocked is the only warehouse-wide half: when true the SP cannot edit this warehouse at all, so no axis can be enabled. A top-level blocked with can_manage_blocked false only means Auto Stop has nothing to apply; Warehouse Sizing may still be enableable. Each axis block's enable_preflight is that axis's own gate.

One non-auto-terminate axis's status block. No current_auto_stop_mins — that is warehouse-level and stays top-level only. enable_preflight IS axis-level: each axis is enableable only while it has something to apply, so the gate has to be reported per axis rather than folded into the warehouse-level preflight.