Skip to content

armgit / Pulse

Public Imported from GitHub
0

Real-time monitoring dashboard for Proxmox VE, PBS, Docker, Kubernetes, TrueNAS and vSphere. Self-hosted, with alerts for failed backups, full disks, unhealthy containers and offline nodes.

Reuse bounded rollup query parameters

rcourtmane9dbca7c2b71
diff --git a/docs/release-control/v6/internal/records/workload-chart-idle-proof-2026-10-18.md b/docs/release-control/v6/internal/records/workload-chart-idle-proof-2026-10-18.md
index 250df28938..bf598bd46c 100644
--- a/docs/release-control/v6/internal/records/workload-chart-idle-proof-2026-10-18.md
+++ b/docs/release-control/v6/internal/records/workload-chart-idle-proof-2026-10-18.md
@@ -193,3 +193,64 @@ medians of 4.735637/5.7315405 ms for the small rollup and
are retained without substituting for the revised paired CI result or the
still-pending strict idle-worker chart proof. The previous executable and
initial adverse CI results are preserved.
+
+## Revised paired result and shared bindings
+
+The revised comparison in
+[run 39177248750](https://github.com/rcourtman/Pulse/actions/runs/39177248750)
+used PR head `fc58c597caf0051389c7bb3a5a6e10e7dfb3ac9b` through synthetic merge
+`e79e0546086124d96ee75bd64722c8fc3f8ed6bf`, tree
+`2143d881c0c8cbcb7d8696c204f42149623a3d5b`. Its other parent was current main
+`6209d03d5a3895005f48b365cc2eeb8f1483f072`; the main advance changed only
+Proxmox guest-test fixtures. The paired baseline remained
+`ed4a2c6b54ee1d139d5f198950c53b403596c17c`.
+
+The small rollup improved from 4.309 to 4.254 ms (-1.28%, p = 0.011), and the
+500-node rollup from 121.2 to 112.1 ms (-7.48%, p < 0.001). The four-hour cold
+chart improved from 113.97 to 78.69 ms (-30.96%, p < 0.001), with unchanged
+21.47 MiB and approximately 508,200 allocations. The remaining benchmark flag
+was the small rollup's 3.117 to 3.453 KiB (+10.76%, p < 0.001). Its allocation
+count rose from 100 to 105. The 500-node case likewise gained five allocations
+and about 352 bytes, below that case's relative flag threshold. Artifact
+[12155027036](https://github.com/rcourtman/Pulse/actions/runs/39177248750/artifacts/12155027036)
+has GitHub digest
+`f238c3230b214064fde984765b81c6f93f22f7addb410f7aad43d5ba404a69cd`.
+
+The duplicated discovery tier/window arguments caused those extra allocations.
+Both stages now reuse the ordinary aggregate's same six SQLite binding slots,
+removing the second argument array and repeated boxing without changing
+resource discovery, primary ranges or aggregation semantics. The existing
+cross-layout aggregate and query-plan tests exercise the shared bindings.
+
+The hosted strict test in that revised run passed its latency assertion:
+p50 50.135919 ms, p95 67.674439 ms and p99 74.205989 ms, explicit SLO 90 ms.
+It still failed capacity validation. Two of 12 concurrent capacity samples
+recorded `provjobd1628928` at 21.0% and 25.25% total CPU. The four-vCPU runner
+was an AMD EPYC 9V45, with GOMAXPROCS=2 and Go 1.26.9. This is different
+hardware from the prior public attempt; their absolute timings are not a
+source-only comparison. Admission/after receipts, both adverse intervals and
+the passing test remain retained. Fixed executable SHA-256:
+`89dfcd17f28ccd16d88d1b663ff8266975486adb5804876ae65c1269a4952b98`.
+Artifact
+[12155196024](https://github.com/rcourtman/Pulse/actions/runs/39177248750/artifacts/12155196024)
+has GitHub digest
+`65d01e1a1614836a025f747869d577daa21032a6a362a621413807d4b2c1e150`.
+The test establishes a passing latency assertion on that runner; the failed
+capacity guard still prevents an idle-worker qualification claim.
+
+The shared-binding executable passed the complete targeted rollup correctness
+and plan suite and the rollup race suite (6.010 s). SHA-256:
+`ce878df8c9df1193b457742ef913dac8073f29777b53aeb9148fecfd8b9a1686`.
+Its five source hashes are retained in `rollup-bindings/source.sha256`.
+Ten native samples per case measured small-rollup median allocation volume
+3,204.5 versus 3,532.5 bytes, and 500-node volume 3,896 versus 4,248 bytes.
+Median allocation counts were 106 versus 105 and 113 versus 112 respectively;
+reused numbered bindings reduce volume while the driver handles their names.
+Sequential shared-worker median times were 7.8832245 versus 6.780017 ms and
+164.7742585 versus 149.3372525 ms. These adverse timing observations are
+retained without attributing them to the binding change or substituting for
+fresh paired CI. The full metrics suite passed before the binding-only change;
+the final native proof exercises all affected rollup paths and boundaries.
+All nine required source checks for `fc58c597caf0051389c7bb3a5a6e10e7dfb3ac9b`
+ultimately passed, including REST-0 at 23:57:41Z. Its allocation flag and
+capacity-invalidated SLO are preserved separately from those source checks.
diff --git a/docs/release-control/v6/internal/subsystems/performance-and-scalability.md b/docs/release-control/v6/internal/subsystems/performance-and-scalability.md
index 87e05ed6e2..15cd7d7fe8 100644
--- a/docs/release-control/v6/internal/subsystems/performance-and-scalability.md
+++ b/docs/release-control/v6/internal/subsystems/performance-and-scalability.md
@@ -2778,7 +2778,10 @@ the current production schema. Clustered rollups first discover resource
identities within the bounded source window through the covering tier/time
index, then read each resource through its primary-key time range. They must
not seek the primary table once per source observation through the secondary
-index. A statement-local resource number gives the aggregate an integer group
+index. Discovery and value reads share the same bound source tier and time
+window. Reuse the ordinary aggregate's binding slots so constructing that
+indexed query does not duplicate argument arrays or boxing per window.
+A statement-local resource number gives the aggregate an integer group
key without changing stored resource identity, bucket boundaries, nullable
extrema fallback or `INSERT OR IGNORE` idempotence. It is neither persisted nor
cached. Legacy rowid databases retain their existing aggregate query until
diff --git a/pkg/metrics/clustered_layout.go b/pkg/metrics/clustered_layout.go
index ee2623e8e6..b96b47ea35 100644
--- a/pkg/metrics/clustered_layout.go
+++ b/pkg/metrics/clustered_layout.go
@@ -22,10 +22,12 @@ const clusteredMetricsIndex = "sqlite_autoindex_metrics_1"
// ranges. Group by a statement-local resource number to avoid comparing both
// identity strings for every input row; the number is unique within this query
// and never becomes a persisted or externally visible resource identity.
+// Reuse the ordinary rollup's six binding slots in both read stages. Repeating
+// the tier/window arguments would allocate another argument slice per window.
const clusteredMetricsRollupSQL = `
WITH selected_resources AS MATERIALIZED (
SELECT DISTINCT resource_type, resource_id FROM metrics
- WHERE tier = ? AND timestamp >= ? AND timestamp < ?
+ WHERE tier = ?4 AND timestamp >= ?5 AND timestamp < ?6
), resources AS MATERIALIZED (
SELECT ROW_NUMBER() OVER (ORDER BY resource_type, resource_id) AS resource_number,
resource_type, resource_id FROM selected_resources
@@ -34,10 +36,10 @@ const clusteredMetricsRollupSQL = `
(resource_type, resource_id, metric_type, value, min_value, max_value, timestamp, tier)
SELECT r.resource_type, r.resource_id, m.metric_type, AVG(m.value),
MIN(COALESCE(m.min_value, m.value)), MAX(COALESCE(m.max_value, m.value)),
- (m.timestamp / ?) * ? AS bucket_ts, ?
+ (m.timestamp / ?1) * ?2 AS bucket_ts, ?3
FROM resources AS r CROSS JOIN metrics AS m
WHERE m.resource_type = r.resource_type AND m.resource_id = r.resource_id
- AND m.tier = ? AND m.timestamp >= ? AND m.timestamp < ?
+ AND m.tier = ?4 AND m.timestamp >= ?5 AND m.timestamp < ?6
GROUP BY r.resource_number, bucket_ts, m.metric_type
ORDER BY r.resource_number, bucket_ts, m.metric_type
`
diff --git a/pkg/metrics/clustered_layout_test.go b/pkg/metrics/clustered_layout_test.go
index d61082f284..86f8c4b0a4 100644
--- a/pkg/metrics/clustered_layout_test.go
+++ b/pkg/metrics/clustered_layout_test.go
@@ -67,7 +67,6 @@ func TestClusteredMetricsRollupPlanUsesBoundedPrimaryRanges(t *testing.T) {
}
t.Cleanup(func() { _ = store.Close() })
rows, err := store.db.Query("EXPLAIN QUERY PLAN "+clusteredMetricsRollupSQL,
- "raw", int64(1_700_000_000), int64(1_700_000_120),
int64(60), int64(60), "minute", "raw", int64(1_700_000_000), int64(1_700_000_120))
if err != nil {
t.Fatal(err)
diff --git a/pkg/metrics/store.go b/pkg/metrics/store.go
index dacc21032f..3d73865290 100644
--- a/pkg/metrics/store.go
+++ b/pkg/metrics/store.go
@@ -2377,7 +2377,6 @@ func (s *Store) rollupTierWindow(fromTier, toTier Tier, bucketSecs, startTs, end
params := []interface{}{bucketSecs, bucketSecs, string(toTier), string(fromTier), startTs, endTs}
if s.clusteredMetrics.Load() {
query = clusteredMetricsRollupSQL
- params = append([]interface{}{string(fromTier), startTs, endTs}, params...)
}
_, err = tx.Exec(query, params...)
if err != nil {