app/vmselect: handle negative time range start in a generic manner inside NewSearchQuery()

This is a follow-up for cf03e11d89

Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/5553
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/pull/5630

app/vmselect/prometheus/prometheus.go

@@ -794,9 +794,6 @@ func QueryHandler(qt *querytracer.Tracer, startTime time.Time, at *auth.Token, w
 		start -= offset
 		end := start
 		start = end - window
-		if start < 0 {
-			start = 0
-		}
 		// Do not include data point with a timestamp matching the lower boundary of the window as Prometheus does.
 		start++
 		if end < start {

app/vmselect/promql/eval.go

@@ -1105,9 +1105,6 @@ func evalRollupFuncWithMetricExpr(qt *querytracer.Tracer, ec *EvalConfig, funcNa
 	} else {
 		minTimestamp -= ec.Step
 	}
-	if minTimestamp < 0 {
-		minTimestamp = 0
-	}
 	sq := storage.NewSearchQuery(ec.AuthToken.AccountID, ec.AuthToken.ProjectID, minTimestamp, ec.End, tfss, ec.MaxSeries)
 	rss, isPartial, err := netstorage.ProcessSearchQuery(qt, ec.DenyPartialResponse, sq, ec.Deadline)
 	if err != nil {

docs/CHANGELOG.md

@@ -11,9 +11,9 @@ The following `tip` changes can be tested by building VictoriaMetrics components
 
 ## v1.93.x long-time support release (LTS)
 
-* BUGFIX: [vmselect](https://docs.victoriametrics.com/vmselect.html): properly determine time range search for instant queries with too big look-behind window like `foo[100y]`. Previously, such queries could return empty responses even if `foo` is present in database.
 * BUGFIX: properly return errors from [export APIs](https://docs.victoriametrics.com/#how-to-export-time-series). Previously these errors were silently suppressed. See [this pull request](https://github.com/VictoriaMetrics/VictoriaMetrics/pull/5649).
 * BUGFIX: [vmagent](https://docs.victoriametrics.com/vmagent.html): properly discover targets for `role: endpoints` and `role: endpointslice` in [kubernetes_sd_configs](https://docs.victoriametrics.com/sd_configs.html#kubernetes_sd_configs). Previously some `endpoints` and `endpointslice` targets could be left undiscovered or some targets could have missing `__meta_*` labels when performing service discovery in busy Kubernetes clusters with large number of pods. See [this pull request](https://github.com/VictoriaMetrics/VictoriaMetrics/pull/5557).
+* BUGFIX: [MetricsQL](https://docs.victoriametrics.com/MetricsQL.html): properly process queries with too big lookbehind window such as `foo[100y]`. Previously, such queries could return empty responses even if `foo` is present in database. See [this issue](https://github.com/VictoriaMetrics/VictoriaMetrics/issues/5553).
 * BUGFIX: [MetricsQL](https://docs.victoriametrics.com/MetricsQL.html): properly handle possible negative results caused by float operations precision error in rollup functions like rate() or increase(). See [this issue](https://github.com/VictoriaMetrics/VictoriaMetrics/issues/5571).
 
 ## [v1.93.10](https://github.com/VictoriaMetrics/VictoriaMetrics/releases/tag/v1.93.10)

lib/storage/search.go

@@ -300,6 +300,10 @@ func (sq *SearchQuery) GetTimeRange() TimeRange {
 
 // NewSearchQuery creates new search query for the given args.
 func NewSearchQuery(accountID, projectID uint32, start, end int64, tagFilterss [][]TagFilter, maxMetrics int) *SearchQuery {
+	if start < 0 {
+		// This is needed for https://github.com/VictoriaMetrics/VictoriaMetrics/issues/5553
+		start = 0
+	}
 	if maxMetrics <= 0 {
 		maxMetrics = 2e9
 	}