Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion PerformanceMonitor.Collectors/PlanCorrectionCollector.cs
Original file line number Diff line number Diff line change
Expand Up @@ -148,7 +148,7 @@ public sealed class Row
///
/// <para>
/// Correcting the record, because it was wrong once and cost a round trip: #2760 removed both hints
/// on the theory that ~4,237ms of avg CPU seen on <host>/<database> was compile cost. That
/// on the theory that ~4,237ms of avg CPU seen on &lt;host&gt;/&lt;database&gt; was compile cost. That
/// attribution was never verified — it was inferred from a CPU-heavy/low-read/low-row telemetry
/// signature. The cost was the LIVE Query Store catalog-view joins that #2764 then replaced with
/// staged temps. Removing RECOMPILE never addressed it; staging did.
Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -65,7 +65,7 @@ public static class QueryStoreOpenIntervalState
/// recorded rationale beats a knob nobody can reason about.
///
/// <para>Moved from 15 to 30 (#2759) because the #2312 yardstick it was tuned against — multi-53 at
/// ~50 s/run — stopped holding. Measured live on <c><host></c> (2026-09-01, 3-hour
/// ~50 s/run — stopped holding. Measured live on <c>&lt;host&gt;</c> (2026-09-01, 3-hour
/// window of <c>collection_log</c>): open-interval-inclusive runs cost 110–133 s each (2–2.5x the
/// original yardstick), landing every 12–19 minutes — the 15-minute skip window was already firing as
/// designed, spacing the expensive runs out, but each one had grown too large for that spacing to be
Expand Down
Loading