fix: drop non-finite values for promql, revert cache change

docs/contributing/go/handler.md

@@ -349,7 +349,7 @@ func (Step) JSONSchema() (jsonschema.Schema, error) {
 
 ### `oneOf` with a discriminator
 
-For a sum type whose variants are keyed by a property (e.g. `kind`), expose the variants via `JSONSchemaOneOf()` and add a discriminator. Without it, code generators intersect the variants (`A & B & C`) instead of producing a clean discriminated union (`A | B | C`). How to model the sum type itself is covered in [types.md](types.md#sum-types-the-kindspec-envelope) — this section is only about its schema.
+For a sum type whose variants are keyed by a property (e.g. `kind`), expose the variants via `JSONSchemaOneOf()` and add a discriminator. Without it, code generators intersect the variants (`A & B & C`) instead of producing a clean discriminated union (`A | B | C`).
 
 The parent keeps its `JSONSchemaOneOf()` (the `oneOf` itself) and *additionally* tags it via `PrepareJSONSchema` with the `x-signoz-discriminator` extension; `signoz.attachDiscriminators` then promotes that marker to a real OpenAPI 3 `discriminator` (and strips the duplicate parent properties) after reflection.
 

docs/contributing/go/types.md

@@ -99,69 +99,6 @@ Each flavor exists for a concrete reason:
 
 The core `AuthDomain` holds the two live halves — `storableAuthDomain` and `authDomainConfig` — and owns business methods such as `Update(config)`. Conversions use the `New<Output>From<Input>` form: `NewAuthDomainFromConfig`, `NewAuthDomainFromStorableAuthDomain`, `NewGettableAuthDomainFromAuthDomain`.
 
-## Sum types: the kind/spec envelope
-
-When a domain type is a *sum type* — exactly one of several variants, selected by a discriminator — model it as an envelope with a `kind` and a `spec`:
-
-```go
-type FooConfig struct {
-    Kind FooKind `json:"kind" required:"true"`
-    Spec any     `json:"spec" required:"true"`
-}
-```
-
-```json
-{ "kind": "bar", "spec": { "url": "...", "timeout": "30s" } }
-```
-
-`Kind` is a `valuer.String` enum implementing `Enum()`; `Spec` holds exactly one concrete variant type (`BarSpec`, `BazSpec`, …). `RuleThresholdData` and `EvaluationEnvelope` in `pkg/types/ruletypes/` are the canonical in-tree examples; the dashboard panel/query/variable plugins in `pkg/types/dashboardtypes/` are the same pattern behind generics. (`QueryEnvelope` in querybuildertypes uses `type` as the discriminator key for historical reasons; new envelopes use `kind`.)
-
-### The envelope goes at the point of variance, not the resource root
-
-Put the envelope on the field that actually varies. The resource root is almost never a sum type — a `Foo` has a `name` and an `enabled` flag regardless of which kind it is configured with; only its configuration varies, so the envelope is the `config` field:
-
-```json
-{ "name": "my-foo", "enabled": true, "config": { "kind": "bar", "spec": { "...": "..." } } }
-```
-
-Hoisting `kind`/`spec` to the root would turn the whole resource into a `oneOf`: every flavor (`PostableFoo`, `UpdatableFoo`, `GettableFoo`) then needs one variant schema per kind, each repeating the common fields; every new common field has to be added to all of them; and generated clients get unions of large objects instead of one small union that narrows on `config.kind`. A root-level `kind` also collides with the resource-model meaning of the word — root `kind` conventionally answers "what resource is this" (`Dashboard`), never "which flavor of config does it hold".
-
-The existing domains already follow this placement:
-
-- **Rules** — plain root; envelopes on the varying fields: `thresholds: {kind, spec}` and `evaluation: {kind, spec}`.
-- **Dashboards** — metadata at the root plus one typed `spec`; the unions sit deep inside, at each panel/query/variable plugin (`{kind, spec}` in `perses_plugin_wrappers.go`).
-- **Saved views** — root `{schemaVersion, spec}`, where `spec` is a *versioning* envelope holding one fixed type, not a union; the unions are inside it (`spec.queries: [{type, spec}]`). Same word, different job — a versioned body is not a discriminated union.
-
-### Why this tagging style
-
-Of the union encodings in common use, the envelope is the *adjacently tagged* one — tag and payload side by side. Variant payloads stay collision-free, and each kind maps to a named wrapper schema that carries the discriminator, which is exactly what OpenAPI generators need. The alternatives lose on those points: *internally tagged* (`{"kind": "bar", ...fields flattened}`) mixes common and variant fields, admits cross-variant key collisions, and forces every variant schema to redeclare the discriminator; *sibling optional fields* (`{"kind": "bar", "barConfig": {}, "bazConfig": {}}`) is the anti-pattern the first rule below exists to prevent.
-
-The rules that make the envelope work:
-
-- **Never model variants as sibling fields.** A struct with `Bar *BarSpec`, `Baz *BazSpec` next to a discriminator cannot be expressed as an OpenAPI discriminated union, forces nilability checks on every consumer, and silently admits contradictory payloads (kind=bar with a baz spec). The chosen variant *is* the payload.
-- **The envelope owns `UnmarshalJSON`.** Decode `kind` first, then switch on it to decode and validate the matching concrete type into `Spec`. Unknown kinds and missing specs are rejected at the boundary:
-
-    ```go
-    func (typ *FooConfig) UnmarshalJSON(data []byte) error {
-        var raw map[string]json.RawMessage
-        // ... unmarshal raw, decode raw["kind"] ...
-        switch kind {
-        case FooKindBar:
-            spec := BarSpec{}
-            if err := json.Unmarshal(raw["spec"], &spec); err != nil {
-                return err
-            }
-            typ.Spec = spec
-        // ... one case per kind, default rejects ...
-        }
-        typ.Kind = kind
-        return nil
-    }
-    ```
-- **Consumers type-assert on `Spec`** (`config.Spec.(BarSpec)`) after switching on `Kind`. If assertion sites multiply, add typed accessors on the envelope (see `EvaluationEnvelope.GetEvaluation()`).
-- **OpenAPI needs one unexported variant struct per kind** (`fooConfigBar{Kind; Spec BarSpec}`), exposed via `JSONSchemaOneOf()` and mapped via `PrepareJSONSchema` with the `x-signoz-discriminator` extension. The schema mechanics are covered in [handler.md](handler.md#oneof-with-a-discriminator).
-- **A legacy persisted shape gets a data migration or a `StorableX`.** When rows were written before the envelope existed, prefer an idempotent `sqlmigration` that rewrites them into the new shape, so the storable type simply nests the envelope. Only when the old shape must keep being written (external writers, rollback windows) keep it in a storable twin and convert at the type boundary.
-
 ## Conventions that tie the flavors together
 
 - **Conversions** use either a `New<Output>From<Input>` constructor — e.g. `NewChannelFromReceiver`, `NewGettableAuthDomainFromAuthDomain` — or a receiver-style `ToY()` method. Both forms coexist in the codebase; use whichever fits the call site.
@@ -202,8 +139,6 @@ Both are optional. Do not introduce them if `PostableX` already covers the case.
 
 - Every domain package defines the core type `X`. Only `X` is mandatory.
 - Add `PostableX` / `GettableX` / `UpdatableX` / `StorableX` one at a time, only when the shape actually diverges from `X`.
-- Model sum types as a `{kind, spec}` envelope with a validating `UnmarshalJSON` — never as sibling variant fields next to a discriminator.
-- The envelope goes on the field that varies, never at the resource root — common fields stay on the resource, outside the union.
 - Domain logic lives on `X`, not on the flavor types.
 - Conversions can be a `New<Output>From<Input>` constructor or a receiver-style `ToY()` method — pick whichever reads best at the call site.
 - Use a type alias when two shapes are truly identical.

frontend/src/components/LogDetail/LogHighlights/LogHighlights.module.scss

@@ -1,46 +0,0 @@
-.highlights {
-	display: grid;
-	grid-template-columns: repeat(4, minmax(0, 1fr));
-	gap: 12px 16px;
-	padding: 12px 0;
-
-	// Constrain each KeyValueLabel (the grid items) to its cell.
-	:global(.key-value-label) {
-		width: auto;
-		min-width: 0;
-		overflow: hidden;
-	}
-}
-
-.valueBadge {
-	--badge-font-size: 13px;
-	box-sizing: border-box;
-	max-width: 100%;
-	min-width: 0;
-}
-
-// Truncating text inside a badge
-.badgeText {
-	min-width: 0;
-	overflow: hidden;
-	text-overflow: ellipsis;
-	white-space: nowrap;
-}
-
-.serviceDot {
-	width: 6px;
-	height: 6px;
-	border-radius: 50%;
-	background: var(--accent-forest);
-	flex-shrink: 0;
-	margin-right: 4px;
-}
-
-.traceLink {
-	display: inline-block;
-	max-width: 100%;
-	overflow: hidden;
-	text-overflow: ellipsis;
-	white-space: nowrap;
-	color: var(--accent-primary);
-}

frontend/src/components/LogDetail/LogHighlights/LogHighlights.tsx

@@ -1,36 +0,0 @@
-import KeyValueLabel from 'periscope/components/KeyValueLabel';
-import { ILog } from 'types/api/logs/log';
-
-import { LOG_HIGHLIGHTS } from './config';
-import styles from './LogHighlights.module.scss';
-
-interface LogHighlightsProps {
-	log: ILog;
-}
-
-function LogHighlights({ log }: LogHighlightsProps): JSX.Element | null {
-	const fields = LOG_HIGHLIGHTS.map((field) => ({
-		key: field.key,
-		label: field.label,
-		value: field.render(log),
-	})).filter((field) => field.value != null);
-
-	if (fields.length === 0) {
-		return null;
-	}
-
-	return (
-		<div className={styles.highlights} data-testid="log-details-highlights">
-			{fields.map((field) => (
-				<KeyValueLabel
-					key={field.key}
-					badgeKey={field.label}
-					badgeValue={field.value}
-					direction="column"
-				/>
-			))}
-		</div>
-	);
-}
-
-export default LogHighlights;

frontend/src/components/LogDetail/LogHighlights/TraceIdField.tsx

@@ -1,23 +0,0 @@
-import { Link } from 'react-router-dom';
-
-import styles from './LogHighlights.module.scss';
-
-interface TraceIdFieldProps {
-	traceId: string;
-}
-
-function TraceIdField({ traceId }: TraceIdFieldProps): JSX.Element {
-	return (
-		<Link
-			to={{ pathname: `/trace/${traceId}` }}
-			target="_blank"
-			rel="noreferrer"
-			className={styles.traceLink}
-			title={traceId}
-		>
-			{traceId}
-		</Link>
-	);
-}
-
-export default TraceIdField;

frontend/src/components/LogDetail/LogHighlights/config.tsx

@@ -1,102 +0,0 @@
-import { ReactNode } from 'react';
-import { Badge, BadgeColor } from '@signozhq/ui/badge';
-import { LogType } from 'components/Logs/LogStateIndicator/LogStateIndicator';
-import { getLogIndicatorType } from 'components/Logs/LogStateIndicator/utils';
-import { ILog } from 'types/api/logs/log';
-
-import styles from './LogHighlights.module.scss';
-import TraceIdField from './TraceIdField';
-
-// Severity badge color mirrors the LogStateIndicator bar
-const SEVERITY_COLOR: Record<string, BadgeColor> = {
-	[LogType.TRACE]: 'forest',
-	[LogType.DEBUG]: 'aqua',
-	[LogType.INFO]: 'robin',
-	[LogType.WARN]: 'amber',
-	[LogType.ERROR]: 'cherry',
-	[LogType.FATAL]: 'sakura',
-};
-
-export interface LogHighlightConfig {
-	key: string;
-	label: string;
-	render: (log: ILog) => ReactNode | null;
-}
-
-// Resource/attribute lookup (keys like `service.name` live in resources_string,
-// occasionally attributes_string). Typed loosely as these are string maps.
-const getAttr = (log: ILog, key: string): string =>
-	(log.resources_string as unknown as Record<string, string>)?.[key] ||
-	(log.attributes_string as unknown as Record<string, string>)?.[key] ||
-	'';
-
-const valueBadge = (
-	value: string,
-	options?: { prefix?: ReactNode; color?: BadgeColor },
-): ReactNode => (
-	<Badge color={options?.color ?? 'vanilla'} className={styles.valueBadge}>
-		{options?.prefix}
-		<span className={styles.badgeText} title={value}>
-			{value}
-		</span>
-	</Badge>
-);
-
-export const LOG_HIGHLIGHTS: LogHighlightConfig[] = [
-	{
-		key: 'service',
-		label: 'SERVICE',
-		render: (log): ReactNode | null => {
-			const value = getAttr(log, 'service.name');
-			return value
-				? valueBadge(value, {
-						prefix: <span className={styles.serviceDot} />,
-					})
-				: null;
-		},
-	},
-	{
-		key: 'severity',
-		label: 'SEVERITY',
-		render: (log): ReactNode | null => {
-			if (!log.severity_text) {
-				return null;
-			}
-			return valueBadge(log.severity_text, {
-				color: SEVERITY_COLOR[getLogIndicatorType(log)] ?? 'vanilla',
-			});
-		},
-	},
-	{
-		key: 'namespace',
-		label: 'NAMESPACE',
-		render: (log): ReactNode | null => {
-			const value = getAttr(log, 'service.namespace');
-			return value ? valueBadge(value) : null;
-		},
-	},
-	{
-		key: 'environment',
-		label: 'ENVIRONMENT',
-		render: (log): ReactNode | null => {
-			const value = getAttr(log, 'deployment.environment');
-			return value ? valueBadge(value) : null;
-		},
-	},
-	{
-		key: 'traceId',
-		label: 'TRACE ID',
-		render: (log): ReactNode | null => {
-			const traceId = log.trace_id || log.traceId;
-			return traceId ? <TraceIdField traceId={traceId} /> : null;
-		},
-	},
-	{
-		key: 'spanId',
-		label: 'SPAN ID',
-		render: (log): ReactNode | null => {
-			const spanId = log.span_id || log.spanID;
-			return spanId ? valueBadge(spanId) : null;
-		},
-	},
-];

frontend/src/components/LogDetail/__tests__/LogDetail.test.tsx

@@ -115,45 +115,6 @@ describe('LogDetail drawer — header (isLogDetailsV2)', () => {
 		expect(screen.queryByText('Open in Explorer')).not.toBeInTheDocument();
 	});
 
-	it('renders Highlights for fields present on the log, omitting absent ones', () => {
-		const logWithMeta = {
-			...mockLog,
-			severity_text: 'ERROR',
-			trace_id: 'trace-abc',
-			resources_string: {
-				'service.name': 'checkout',
-				'deployment.environment': 'production',
-			},
-		} as unknown as ILog;
-
-		renderDrawer({ log: logWithMeta });
-
-		const highlights = screen.getByTestId('log-details-highlights');
-		expect(highlights).toHaveTextContent('SEVERITY');
-		expect(highlights).toHaveTextContent('ERROR');
-		expect(highlights).toHaveTextContent('SERVICE');
-		expect(highlights).toHaveTextContent('checkout');
-		expect(highlights).toHaveTextContent('ENVIRONMENT');
-		expect(highlights).toHaveTextContent('production');
-		expect(highlights).toHaveTextContent('TRACE ID');
-		// Absent fields are omitted (no namespace / span id on this log).
-		expect(highlights).not.toHaveTextContent('NAMESPACE');
-		expect(highlights).not.toHaveTextContent('SPAN ID');
-	});
-
-	it('links the trace id highlight to the trace detail in a new tab', () => {
-		const logWithTrace = {
-			...mockLog,
-			trace_id: 'trace-abc',
-		} as unknown as ILog;
-
-		renderDrawer({ log: logWithTrace });
-
-		const link = screen.getByRole('link', { name: 'trace-abc' });
-		expect(link).toHaveAttribute('target', '_blank');
-		expect(link.getAttribute('href')).toContain('/trace/trace-abc');
-	});
-
 	it('navigates to the next / previous log with the Down / Up arrow keys', async () => {
 		const user = userEvent.setup({ pointerEventsCheck: 0 });
 		const logs = [makeLog('log-0'), makeLog('log-1'), makeLog('log-2')];

frontend/src/components/LogDetail/index.tsx

@@ -55,7 +55,6 @@ import { isLogDetailsV2, RESOURCE_KEYS, VIEW_TYPES, VIEWS } from './constants';
 import { LogDetailInnerProps, LogDetailProps } from './LogDetail.interfaces';
 import LogDetailsHeader from './LogDetailsHeader/LogDetailsHeader';
 import { useLogNavigation } from './LogDetailsHeader/useLogNavigation';
-import LogHighlights from './LogHighlights/LogHighlights';
 
 import './LogDetails.styles.scss';
 
@@ -400,8 +399,6 @@ function LogDetailInner({
 					<div className="log-overflow-shadow">&nbsp;</div>
 				</div>
 
-				{isLogDetailsV2 && <LogHighlights log={log} />}
-
 				<div className="tabs-and-search">
 					<ToggleGroupSimple
 						type="single"

frontend/src/components/QueryBuilderV2/QueryV2/QuerySearch/QuerySearch.tsx

@@ -183,14 +183,15 @@ function QuerySearch({
 				isProgrammaticChangeRef.current = true;
 			}
 
-			const changes = view.state.changes({
-				from: 0,
-				to: currentValue.length,
-				insert: value,
-			});
 			view.dispatch({
-				changes,
-				selection: { anchor: changes.newLength },
+				changes: {
+					from: 0,
+					to: currentValue.length,
+					insert: value,
+				},
+				selection: {
+					anchor: value.length,
+				},
 			});
 		},
 		[],

frontend/src/components/QueryBuilderV2/QueryV2/__tests__/QuerySearch.test.tsx

@@ -301,66 +301,6 @@ describe('QuerySearch (Integration with Real CodeMirror)', () => {
 		dispatchSpy.mockRestore();
 	});
 
-	it('does not crash when the expression contains CRLF line breaks (issue #5869)', async () => {
-		const dispatchSpy = jest.spyOn(EditorView.prototype, 'dispatch');
-		const onChange = jest.fn() as jest.MockedFunction<(v: string) => void>;
-		const initialExpression = "service.name = 'frontend'";
-		// Filtering on a multi-line log value (CRLF) used to throw
-		// "RangeError: Selection points outside of document".
-		const crlfExpression = "body CONTAINS 'line1\r\nline2\r\nline3'";
-
-		const baseQueryData = {
-			...initialQueriesMap.logs.builder.queryData[0],
-			filter: { expression: initialExpression },
-		};
-
-		const { rerender } = render(
-			<QuerySearch
-				onChange={onChange}
-				queryData={baseQueryData}
-				dataSource={DataSource.LOGS}
-			/>,
-		);
-
-		await waitFor(
-			() => {
-				const editorContent = document.querySelector(
-					CM_EDITOR_SELECTOR,
-				) as HTMLElement;
-				expect(editorContent.textContent || '').toBe(initialExpression);
-			},
-			{ timeout: 3000 },
-		);
-
-		rerender(
-			<QuerySearch
-				onChange={onChange}
-				queryData={{ ...baseQueryData, filter: { expression: crlfExpression } }}
-				dataSource={DataSource.LOGS}
-			/>,
-		);
-
-		// The programmatic replace dispatched without throwing, and the selection anchor
-		// stayed within the CRLF-normalized document (the bug set it past the end).
-		await waitFor(() => {
-			const spec = dispatchSpy.mock.calls
-				.map(
-					(call) =>
-						call[0] as {
-							selection?: { anchor?: number };
-							changes?: { newLength?: number };
-						},
-				)
-				.find((s) => s?.selection?.anchor != null && s?.changes?.newLength != null);
-			expect(spec).toBeDefined();
-			expect(spec?.selection?.anchor).toBeLessThanOrEqual(
-				spec?.changes?.newLength as number,
-			);
-		});
-
-		dispatchSpy.mockRestore();
-	});
-
 	it('fetches key suggestions for metrics even without aggregateAttribute.key when showFilterSuggestionsWithoutMetric is true', async () => {
 		const mockedGetKeys = getKeySuggestions as jest.MockedFunction<
 			typeof getKeySuggestions

pkg/querier/promql_query.go

@@ -5,6 +5,7 @@ import (
 	"context"
 	"fmt"
 	"log/slog"
+	"math"
 	"regexp"
 	"sort"
 	"strings"
@@ -478,11 +479,19 @@ func (q *promqlQuery) toResult(matrix promql.Matrix, warnings []string, began ti
 
 		for idx := range v.Floats {
 			p := v.Floats[idx]
+			// NaN and +/-Inf have no JSON number form and nothing to plot; the
+			// builder path drops them while scanning rows (see consume.go).
+			if math.IsNaN(p.F) || math.IsInf(p.F, 0) {
+				continue
+			}
 			s.Values = append(s.Values, &qbv5.TimeSeriesValue{
 				Timestamp: p.T,
 				Value:     p.F,
 			})
 		}
+		if len(s.Values) == 0 {
+			continue
+		}
 		series = append(series, &s)
 	}
 
@@ -494,13 +503,11 @@ func (q *promqlQuery) toResult(matrix promql.Matrix, warnings []string, began ti
 	}
 	statsMu.Unlock()
 
-	tsData := &qbv5.TimeSeriesData{
-		QueryName: q.query.Name,
-		Aggregations: []*qbv5.AggregationBucket{
-			{
-				Series: series,
-			},
-		},
+	tsData := &qbv5.TimeSeriesData{QueryName: q.query.Name}
+	// No bucket at all when nothing survived: a bucket holding no series reads
+	// as "filtered to empty" to the cache, which stores it as a real result.
+	if len(series) > 0 {
+		tsData.Aggregations = []*qbv5.AggregationBucket{{Series: series}}
 	}
 
 	var payload any = tsData

pkg/querier/promql_query_test.go

@@ -2,14 +2,21 @@ package querier
 
 import (
 	"log/slog"
+	"math"
 	"strings"
+	"sync"
 	"testing"
+	"time"
+
+	"github.com/prometheus/prometheus/model/labels"
+	"github.com/prometheus/prometheus/promql"
 
 	"github.com/SigNoz/signoz/pkg/errors"
 	"github.com/SigNoz/signoz/pkg/prometheus"
 	"github.com/SigNoz/signoz/pkg/prometheus/prometheustest"
 	qbv5 "github.com/SigNoz/signoz/pkg/types/querybuildertypes/querybuildertypesv5"
 	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
 )
 
 func TestRemoveAllVarMatchers(t *testing.T) {
@@ -453,3 +460,82 @@ func TestFingerprint_PinnedProviderBypassesCache(t *testing.T) {
 	}
 	assert.Empty(t, q.Fingerprint())
 }
+
+func TestToResultDropsNonFiniteValues(t *testing.T) {
+	tests := []struct {
+		description        string
+		floats             []promql.FPoint
+		expectedTimestamps []int64
+		expectedValues     []float64
+	}{
+		{
+			description:        "finite values pass through untouched",
+			floats:             []promql.FPoint{{T: 1000, F: 1.5}, {T: 2000, F: 2.5}},
+			expectedTimestamps: []int64{1000, 2000},
+			expectedValues:     []float64{1.5, 2.5},
+		},
+		{
+			description:        "a ratio's 0/0 points are dropped, the rest kept",
+			floats:             []promql.FPoint{{T: 1000, F: 1.5}, {T: 2000, F: math.NaN()}, {T: 3000, F: 2.5}},
+			expectedTimestamps: []int64{1000, 3000},
+			expectedValues:     []float64{1.5, 2.5},
+		},
+		{
+			description:        "both infinities are dropped",
+			floats:             []promql.FPoint{{T: 1000, F: math.Inf(1)}, {T: 2000, F: 4.5}, {T: 3000, F: math.Inf(-1)}},
+			expectedTimestamps: []int64{2000},
+			expectedValues:     []float64{4.5},
+		},
+	}
+
+	for _, test := range tests {
+		t.Run(test.description, func(t *testing.T) {
+			q := &promqlQuery{query: qbv5.PromQuery{Name: "A"}, requestType: qbv5.RequestTypeTimeSeries}
+			matrix := promql.Matrix{{Metric: labels.FromStrings("job_name", "dbBloatMonitorJob"), Floats: test.floats}}
+
+			var mu sync.Mutex
+			var rows, bytes uint64
+			result := q.toResult(matrix, nil, time.Now(), &mu, &rows, &bytes)
+
+			tsData, ok := result.Value.(*qbv5.TimeSeriesData)
+			require.True(t, ok)
+			require.Len(t, tsData.Aggregations, 1)
+			require.Len(t, tsData.Aggregations[0].Series, 1)
+
+			timestamps := make([]int64, 0, len(test.expectedTimestamps))
+			values := make([]float64, 0, len(test.expectedValues))
+			for _, v := range tsData.Aggregations[0].Series[0].Values {
+				timestamps = append(timestamps, v.Timestamp)
+				values = append(values, v.Value)
+			}
+			assert.Equal(t, test.expectedTimestamps, timestamps)
+			assert.Equal(t, test.expectedValues, values)
+		})
+	}
+}
+
+// A series left with nothing must not surface as an empty series, and a result
+// left with no series must carry no aggregation bucket at all — the cache reads
+// a bucket holding no series as a real, filtered-to-empty result and stores it.
+func TestToResultDropsSeriesAndBucketLeftEmpty(t *testing.T) {
+	q := &promqlQuery{query: qbv5.PromQuery{Name: "A"}, requestType: qbv5.RequestTypeTimeSeries}
+	matrix := promql.Matrix{
+		{Metric: labels.FromStrings("job_name", "idleJob"), Floats: []promql.FPoint{{T: 1000, F: math.NaN()}}},
+		{Metric: labels.FromStrings("job_name", "activeJob"), Floats: []promql.FPoint{{T: 1000, F: 7.5}}},
+	}
+
+	var mu sync.Mutex
+	var rows, bytes uint64
+	tsData, ok := q.toResult(matrix, nil, time.Now(), &mu, &rows, &bytes).Value.(*qbv5.TimeSeriesData)
+	require.True(t, ok)
+	require.Len(t, tsData.Aggregations, 1)
+	require.Len(t, tsData.Aggregations[0].Series, 1, "the all-NaN series is gone")
+	assert.Equal(t, "activeJob", tsData.Aggregations[0].Series[0].Labels[0].Value)
+
+	allNaN := promql.Matrix{
+		{Metric: labels.FromStrings("job_name", "idleJob"), Floats: []promql.FPoint{{T: 1000, F: math.NaN()}}},
+	}
+	tsData, ok = q.toResult(allNaN, nil, time.Now(), &mu, &rows, &bytes).Value.(*qbv5.TimeSeriesData)
+	require.True(t, ok)
+	assert.Empty(t, tsData.Aggregations)
+}

tests/integration/tests/promqlconformance/03_nonfinite_cache.py

@@ -0,0 +1,63 @@
+from collections.abc import Callable
+from datetime import UTC, datetime, timedelta
+from http import HTTPStatus
+
+from fixtures import types
+from fixtures.auth import USER_ADMIN_EMAIL, USER_ADMIN_PASSWORD
+from fixtures.metrics import Metrics
+from fixtures.querier import get_all_series, make_query_request
+
+SUM_METRIC = "job_duration_sum"
+COUNT_METRIC = "job_duration_count"
+HOUR_MS = 3_600_000
+SAMPLE_INTERVAL_MS = 60_000
+
+
+def test_promql_ratio_with_zero_denominator_is_dropped_and_cached(
+    signoz: types.SigNoz,
+    create_user_admin: None,  # pylint: disable=unused-argument
+    get_token: Callable[[str, str], str],
+    insert_metrics: Callable[[list[Metrics]], None],
+) -> None:
+    # 12h ending on an hour boundary 15m ago — old enough to be cached.
+    end_ms = (int((datetime.now(tz=UTC) - timedelta(minutes=15)).timestamp() * 1000) // HOUR_MS) * HOUR_MS
+    start_ms = end_ms - 12 * HOUR_MS
+
+    # active_job divides finite; idle_job is 0/0 at every step.
+    series = {"active_job": (100.0, 4.0), "idle_job": (0.0, 0.0)}
+    metrics: list[Metrics] = []
+    for job_name, (sum_value, count_value) in series.items():
+        for ts_ms in range(start_ms, end_ms + 1, SAMPLE_INTERVAL_MS):
+            timestamp = datetime.fromtimestamp(ts_ms / 1000, tz=UTC)
+            metrics.append(Metrics(metric_name=SUM_METRIC, labels={"job_name": job_name}, timestamp=timestamp, value=sum_value))
+            metrics.append(Metrics(metric_name=COUNT_METRIC, labels={"job_name": job_name}, timestamp=timestamp, value=count_value))
+    insert_metrics(metrics)
+
+    token = get_token(USER_ADMIN_EMAIL, USER_ADMIN_PASSWORD)
+    promql = f"sum by (job_name) ({SUM_METRIC}) / sum by (job_name) ({COUNT_METRIC})"
+
+    def run() -> tuple[dict[str, dict[int, object]], int]:
+        query = {"type": "promql", "spec": {"name": "A", "query": promql}}
+        response = make_query_request(signoz, token, start_ms, end_ms, [query], no_cache=False)
+        assert response.status_code == HTTPStatus.OK, response.text[:300]
+        body = response.json()
+        out: dict[str, dict[int, object]] = {}
+        for entry in get_all_series(body, "A") or []:
+            labels = {l["key"]["name"]: str(l["value"]) for l in entry.get("labels") or []}
+            out[labels["job_name"]] = {v["timestamp"]: v["value"] for v in entry.get("values") or []}
+        return out, int(body["data"]["meta"]["stepIntervals"]["A"])
+
+    # First populates the cache, second must be served from it.
+    first, step_seconds = run()
+    second, _ = run()
+
+    expected_points = (end_ms - start_ms) // (step_seconds * 1000) + 1
+    assert set(first) == {"active_job"}, f"the 0/0 series must not reach the response: {sorted(first)}"
+    assert set(first["active_job"].values()) == {25.0}, sorted(set(first["active_job"].values()))
+    assert len(first["active_job"]) == expected_points, f"expected {expected_points} points, got {len(first['active_job'])}"
+
+    # The cached read excludes end_ms, the one legitimate difference.
+    assert set(second) == set(first), sorted(second)
+    for job_name, points in first.items():
+        expected = {ts: value for ts, value in points.items() if ts < end_ms}
+        assert second[job_name] == expected, f"{job_name}: got {len(second[job_name])} of {len(expected)} points"