chore(promql): add clickhousev2 provider (push agg to CH) and shadow-compare promql behind a flag

.github/workflows/integrationci.yaml

@@ -58,6 +58,7 @@ jobs:
           - rootuser
           - serviceaccount
           - querier_json_body
+          - promqlparity
           - querier_skip_resource_fingerprint
           - ttl
         sqlstore-provider:

docs/api/openapi.yml

@@ -1494,7 +1494,7 @@ components:
       - cosmosdb
       - cassandradb
       - redis
-      - cloudsql_postgres
+      - cloudsql
       type: string
     CloudintegrationtypesServiceMetadata:
       properties:
@@ -7998,15 +7998,6 @@ components:
       required:
       - items
       type: object
-    SpantypesGettableSpanMappers:
-      properties:
-        items:
-          items:
-            $ref: '#/components/schemas/SpantypesSpanMapper'
-          type: array
-      required:
-      - items
-      type: object
     SpantypesGettableTraceAggregations:
       properties:
         aggregations:
@@ -8159,7 +8150,7 @@ components:
           type: boolean
         fieldContext:
           $ref: '#/components/schemas/SpantypesFieldContext'
-        groupId:
+        group_id:
           type: string
         id:
           type: string
@@ -8172,7 +8163,7 @@ components:
           type: string
       required:
       - id
-      - groupId
+      - group_id
       - name
       - fieldContext
       - config
@@ -13751,7 +13742,7 @@ paths:
               schema:
                 properties:
                   data:
-                    $ref: '#/components/schemas/SpantypesGettableSpanMappers'
+                    $ref: '#/components/schemas/SpantypesGettableSpanMapperGroups'
                   status:
                     type: string
                 required:

frontend/__mocks__/motionMock.tsx

@@ -1,88 +0,0 @@
-// oxlint-disable-next-line no-restricted-imports
-import * as React from 'react';
-
-// In jsdom, AnimatePresence from motion/react keeps children in DOM during exit
-// animations (awaiting rAF-driven completion that never fully runs in jsdom).
-// This mock makes AnimatePresence render children immediately and makes motion.*
-// elements render as their plain HTML equivalents without animation side-effects.
-//
-// IMPORTANT: motion component references are cached so React sees a stable
-// component identity across re-renders and does not enter an infinite remount loop.
-
-const MOTION_PROPS_TO_STRIP = new Set([
-	'initial',
-	'animate',
-	'exit',
-	'variants',
-	'transition',
-	'whileHover',
-	'whileTap',
-	'whileFocus',
-	'whileInView',
-	'layout',
-	'layoutId',
-	'onAnimationStart',
-	'onAnimationComplete',
-]);
-
-const cache = new Map<string, React.ComponentType>();
-
-function getMotionComponent(tag: string): React.ComponentType {
-	if (!cache.has(tag)) {
-		const Component = React.forwardRef<HTMLElement, Record<string, unknown>>(
-			(props, ref) => {
-				const domProps: Record<string, unknown> = {};
-				for (const [k, v] of Object.entries(props)) {
-					if (!MOTION_PROPS_TO_STRIP.has(k)) {
-						domProps[k] = v;
-					}
-				}
-				return React.createElement(tag, { ...domProps, ref });
-			},
-		);
-		Component.displayName = `motion.${tag}`;
-		cache.set(tag, Component as unknown as React.ComponentType);
-	}
-	return cache.get(tag) as React.ComponentType;
-}
-
-const motionHandler: ProxyHandler<Record<string, React.ComponentType>> = {
-	get(_target, prop: string) {
-		return getMotionComponent(prop);
-	},
-};
-
-export const AnimatePresence: React.FC<{
-	children?: React.ReactNode;
-	mode?: string;
-}> = ({ children }) => React.createElement(React.Fragment, null, children);
-
-export const motion = new Proxy(
-	{} as Record<string, React.ComponentType>,
-	motionHandler,
-);
-
-export const useAnimation = (): Record<string, unknown> => ({
-	start: (): unknown => Promise.resolve(),
-	stop: (): unknown => undefined,
-	set: (): unknown => undefined,
-});
-
-export const useMotionValue = (
-	initial: unknown,
-): { get: () => unknown; set: () => void } => ({
-	get: (): unknown => initial,
-	set: (): unknown => undefined,
-});
-
-export const useTransform = (): { get: () => number } => ({
-	get: (): number => 0,
-});
-
-export const useSpring = (v: unknown): unknown => v;
-
-export const useScroll = (): { scrollY: { get: () => number } } => ({
-	scrollY: { get: (): number => 0 },
-});
-
-export default { motion, AnimatePresence };

frontend/jest.config.ts

@@ -20,7 +20,6 @@ const config: Config.InitialOptions = {
 		'\\.module\\.mjs$': '<rootDir>/__mocks__/cssMock.ts',
 		'\\.md$': '<rootDir>/__mocks__/cssMock.ts',
 		'^uplot$': '<rootDir>/__mocks__/uplotMock.ts',
-		'^motion/react$': '<rootDir>/__mocks__/motionMock.tsx',
 		'^@signozhq/resizable$': '<rootDir>/__mocks__/resizableMock.tsx',
 		'^hooks/useSafeNavigate$': USE_SAFE_NAVIGATE_MOCK_PATH,
 		'^src/hooks/useSafeNavigate$': USE_SAFE_NAVIGATE_MOCK_PATH,

frontend/src/api/generated/services/sigNoz.schemas.ts

@@ -2813,7 +2813,7 @@ export enum CloudintegrationtypesServiceIDDTO {
 	cosmosdb = 'cosmosdb',
 	cassandradb = 'cassandradb',
 	redis = 'redis',
-	cloudsql_postgres = 'cloudsql_postgres',
+	cloudsql = 'cloudsql',
 }
 export type CloudintegrationtypesCloudIntegrationServiceDTOAnyOf = {
 	/**
@@ -9194,76 +9194,6 @@ export interface SpantypesGettableSpanMapperGroupsDTO {
 	items: SpantypesSpanMapperGroupDTO[];
 }
 
-export enum SpantypesSpanMapperOperationDTO {
-	move = 'move',
-	copy = 'copy',
-}
-export interface SpantypesSpanMapperSourceDTO {
-	context: SpantypesFieldContextDTO;
-	/**
-	 * @type string
-	 */
-	key: string;
-	operation: SpantypesSpanMapperOperationDTO;
-	/**
-	 * @type integer
-	 */
-	priority: number;
-}
-
-export interface SpantypesSpanMapperConfigDTO {
-	/**
-	 * @type array,null
-	 */
-	sources: SpantypesSpanMapperSourceDTO[] | null;
-}
-
-export interface SpantypesSpanMapperDTO {
-	config: SpantypesSpanMapperConfigDTO;
-	/**
-	 * @type string
-	 * @format date-time
-	 */
-	createdAt?: string;
-	/**
-	 * @type string
-	 */
-	createdBy?: string;
-	/**
-	 * @type boolean
-	 */
-	enabled: boolean;
-	fieldContext: SpantypesFieldContextDTO;
-	/**
-	 * @type string
-	 */
-	groupId: string;
-	/**
-	 * @type string
-	 */
-	id: string;
-	/**
-	 * @type string
-	 */
-	name: string;
-	/**
-	 * @type string
-	 * @format date-time
-	 */
-	updatedAt?: string;
-	/**
-	 * @type string
-	 */
-	updatedBy?: string;
-}
-
-export interface SpantypesGettableSpanMappersDTO {
-	/**
-	 * @type array
-	 */
-	items: SpantypesSpanMapperDTO[];
-}
-
 export enum SpantypesSpanAggregationTypeDTO {
 	span_count = 'span_count',
 	execution_time_percentage = 'execution_time_percentage',
@@ -9510,6 +9440,30 @@ export interface SpantypesPostableFlamegraphDTO {
 	selectedSpanId?: string;
 }
 
+export enum SpantypesSpanMapperOperationDTO {
+	move = 'move',
+	copy = 'copy',
+}
+export interface SpantypesSpanMapperSourceDTO {
+	context: SpantypesFieldContextDTO;
+	/**
+	 * @type string
+	 */
+	key: string;
+	operation: SpantypesSpanMapperOperationDTO;
+	/**
+	 * @type integer
+	 */
+	priority: number;
+}
+
+export interface SpantypesSpanMapperConfigDTO {
+	/**
+	 * @type array,null
+	 */
+	sources: SpantypesSpanMapperSourceDTO[] | null;
+}
+
 export interface SpantypesPostableSpanMapperDTO {
 	config: SpantypesSpanMapperConfigDTO;
 	/**
@@ -9558,6 +9512,45 @@ export interface SpantypesPostableWaterfallDTO {
 	uncollapsedSpans?: string[] | null;
 }
 
+export interface SpantypesSpanMapperDTO {
+	config: SpantypesSpanMapperConfigDTO;
+	/**
+	 * @type string
+	 * @format date-time
+	 */
+	createdAt?: string;
+	/**
+	 * @type string
+	 */
+	createdBy?: string;
+	/**
+	 * @type boolean
+	 */
+	enabled: boolean;
+	fieldContext: SpantypesFieldContextDTO;
+	/**
+	 * @type string
+	 */
+	group_id: string;
+	/**
+	 * @type string
+	 */
+	id: string;
+	/**
+	 * @type string
+	 */
+	name: string;
+	/**
+	 * @type string
+	 * @format date-time
+	 */
+	updatedAt?: string;
+	/**
+	 * @type string
+	 */
+	updatedBy?: string;
+}
+
 export interface SpantypesUpdatableSpanMapperDTO {
 	config?: SpantypesSpanMapperConfigDTO;
 	/**
@@ -10859,7 +10852,7 @@ export type ListSpanMappersPathParameters = {
 	groupId: string;
 };
 export type ListSpanMappers200 = {
-	data: SpantypesGettableSpanMappersDTO;
+	data: SpantypesGettableSpanMapperGroupsDTO;
 	/**
 	 * @type string
 	 */

frontend/src/components/CreateServiceAccountModal/CreateServiceAccountModal.tsx

@@ -92,7 +92,6 @@ function CreateServiceAccountModal(): JSX.Element {
 			width="narrow"
 			className="create-sa-modal"
 			disableOutsideClick={isErrorModalVisible}
-			testId="create-service-account-modal"
 		>
 			<div className="create-sa-modal__content">
 				<form

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

@@ -1,7 +1,13 @@
 import { toast } from '@signozhq/ui/sonner';
 import { rest, server } from 'mocks-server/server';
 import { NuqsTestingAdapter } from 'nuqs/adapters/testing';
-import { render, screen, userEvent, waitFor } from 'tests/test-utils';
+import {
+	render,
+	screen,
+	userEvent,
+	waitFor,
+	waitForElementToBeRemoved,
+} from 'tests/test-utils';
 
 import CreateServiceAccountModal from '../CreateServiceAccountModal';
 
@@ -83,7 +89,7 @@ describe('CreateServiceAccountModal', () => {
 
 		await waitFor(() => {
 			expect(
-				screen.queryByTestId('create-service-account-modal'),
+				screen.queryByRole('dialog', { name: /New Service Account/i }),
 			).not.toBeInTheDocument();
 		});
 	});
@@ -123,7 +129,7 @@ describe('CreateServiceAccountModal', () => {
 		});
 
 		expect(
-			screen.getByTestId('create-service-account-modal'),
+			screen.getByRole('dialog', { name: /New Service Account/i }),
 		).toBeInTheDocument();
 	});
 
@@ -131,14 +137,15 @@ describe('CreateServiceAccountModal', () => {
 		const user = userEvent.setup({ pointerEventsCheck: 0 });
 		renderModal();
 
-		await screen.findByTestId('create-service-account-modal');
+		const dialog = await screen.findByRole('dialog', {
+			name: /New Service Account/i,
+		});
 		await user.click(screen.getByRole('button', { name: /Cancel/i }));
 
-		await waitFor(() => {
-			expect(
-				screen.queryByTestId('create-service-account-modal'),
-			).not.toBeInTheDocument();
-		});
+		await waitForElementToBeRemoved(dialog);
+		expect(
+			screen.queryByRole('dialog', { name: /New Service Account/i }),
+		).not.toBeInTheDocument();
 	});
 
 	it('shows "Name is required" after clearing the name field', async () => {

frontend/src/components/DownloadOptionsMenu/DownloadOptionsMenu.tsx

@@ -3,7 +3,7 @@ import { Button, Popover, Tooltip } from 'antd';
 import { RadioGroup, RadioGroupItem } from '@signozhq/ui/radio-group';
 import { Typography } from '@signozhq/ui/typography';
 import { TelemetryFieldKey } from 'api/v5/v5';
-import { useExportRawData } from 'hooks/useExportData/useServerExport';
+import { useExportRawData } from 'hooks/useDownloadOptionsMenu/useDownloadOptionsMenu';
 import { Download, LoaderCircle } from '@signozhq/icons';
 import { DataSource } from 'types/common/queryBuilder';
 

frontend/src/components/ServiceAccountDrawer/AddKeyModal/index.tsx

@@ -151,7 +151,6 @@ function AddKeyModal(): JSX.Element {
 			className="add-key-modal"
 			showCloseButton
 			disableOutsideClick={isErrorModalVisible}
-			testId="add-key-modal"
 		>
 			{phase === Phase.FORM && (
 				<KeyFormPhase

frontend/src/components/ServiceAccountDrawer/DeleteAccountModal.tsx

@@ -91,7 +91,6 @@ function DeleteAccountModal(): JSX.Element {
 				color="destructive"
 				loading={isDeleting}
 				onClick={handleConfirm}
-				data-testid="confirm-delete-btn"
 			>
 				<Trash2 size={12} />
 				Delete
@@ -112,7 +111,6 @@ function DeleteAccountModal(): JSX.Element {
 			className="alert-dialog sa-delete-dialog"
 			showCloseButton={false}
 			disableOutsideClick={isErrorModalVisible}
-			testId="delete-service-account-modal"
 			footer={footer}
 		>
 			{content}

frontend/src/components/ServiceAccountDrawer/EditKeyModal/index.tsx

@@ -175,7 +175,6 @@ function EditKeyModal({ keyItem }: EditKeyModalProps): JSX.Element {
 			}
 			showCloseButton={!isRevokeConfirmOpen}
 			disableOutsideClick={isErrorModalVisible}
-			testId="edit-key-modal"
 			footer={
 				isRevokeConfirmOpen ? (
 					<RevokeKeyFooter

frontend/src/components/ServiceAccountDrawer/__tests__/AddKeyModal.test.tsx

@@ -2,7 +2,13 @@ import { toast } from '@signozhq/ui/sonner';
 import { setupAuthzAdmin } from 'lib/authz/utils/authz-test-utils';
 import { rest, server } from 'mocks-server/server';
 import { NuqsTestingAdapter } from 'nuqs/adapters/testing';
-import { render, screen, userEvent, waitFor } from 'tests/test-utils';
+import {
+	render,
+	screen,
+	userEvent,
+	waitFor,
+	waitForElementToBeRemoved,
+} from 'tests/test-utils';
 
 import AddKeyModal from '../AddKeyModal';
 
@@ -91,7 +97,7 @@ describe('AddKeyModal', () => {
 
 		await screen.findByText('snz_abc123xyz456secret');
 		expect(screen.getByText(/Store the key securely/i)).toBeInTheDocument();
-		expect(screen.getByTestId('add-key-modal')).toBeInTheDocument();
+		await screen.findByRole('dialog', { name: /Key Created Successfully/i });
 	});
 
 	it('copy button writes key to clipboard and shows toast.success', async () => {
@@ -127,11 +133,9 @@ describe('AddKeyModal', () => {
 		const user = userEvent.setup({ pointerEventsCheck: 0 });
 		renderModal();
 
-		await screen.findByTestId('add-key-modal');
+		const dialog = await screen.findByRole('dialog', { name: /Add a New Key/i });
 		await user.click(screen.getByRole('button', { name: /Cancel/i }));
 
-		await waitFor(() => {
-			expect(screen.queryByTestId('add-key-modal')).not.toBeInTheDocument();
-		});
+		await waitForElementToBeRemoved(dialog);
 	});
 });

frontend/src/components/ServiceAccountDrawer/__tests__/EditKeyModal.test.tsx

@@ -73,7 +73,9 @@ describe('EditKeyModal (URL-controlled)', () => {
 	it('renders nothing when edit-key param is absent', () => {
 		renderModal(null, { account: 'sa-1' });
 
-		expect(screen.queryByTestId('edit-key-modal')).not.toBeInTheDocument();
+		expect(
+			screen.queryByRole('dialog', { name: /Edit Key Details/i }),
+		).not.toBeInTheDocument();
 	});
 
 	it('renders key data from prop when edit-key param is set', async () => {
@@ -100,7 +102,9 @@ describe('EditKeyModal (URL-controlled)', () => {
 		});
 
 		await waitFor(() => {
-			expect(screen.queryByTestId('edit-key-modal')).not.toBeInTheDocument();
+			expect(
+				screen.queryByRole('dialog', { name: /Edit Key Details/i }),
+			).not.toBeInTheDocument();
 		});
 	});
 
@@ -127,7 +131,9 @@ describe('EditKeyModal (URL-controlled)', () => {
 		expect(latestUrlUpdate.queryString).not.toContain('edit-key=');
 
 		await waitFor(() => {
-			expect(screen.queryByTestId('edit-key-modal')).not.toBeInTheDocument();
+			expect(
+				screen.queryByRole('dialog', { name: /Edit Key Details/i }),
+			).not.toBeInTheDocument();
 		});
 	});
 
@@ -139,7 +145,9 @@ describe('EditKeyModal (URL-controlled)', () => {
 		await user.click(screen.getByRole('button', { name: /Revoke Key/i }));
 
 		// Same dialog, now showing revoke confirmation
-		expect(screen.getByTestId('edit-key-modal')).toBeInTheDocument();
+		await expect(
+			screen.findByRole('dialog', { name: /Revoke Original Key Name/i }),
+		).resolves.toBeInTheDocument();
 		expect(
 			screen.getByText(/Revoking this key will permanently invalidate it/i),
 		).toBeInTheDocument();
@@ -162,7 +170,9 @@ describe('EditKeyModal (URL-controlled)', () => {
 		});
 
 		await waitFor(() => {
-			expect(screen.queryByTestId('edit-key-modal')).not.toBeInTheDocument();
+			expect(
+				screen.queryByRole('dialog', { name: /Edit Key Details/i }),
+			).not.toBeInTheDocument();
 		});
 	});
 });

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

@@ -222,20 +222,21 @@ describe('ServiceAccountDrawer', () => {
 			screen.getByRole('button', { name: /Delete Service Account/i }),
 		);
 
-		await screen.findByTestId('delete-service-account-modal');
-		expect(
-			screen.getByTestId('delete-service-account-modal'),
-		).toBeInTheDocument();
+		const dialog = await screen.findByRole('dialog', {
+			name: /Delete service account CI Bot/i,
+		});
+		expect(dialog).toBeInTheDocument();
 
-		await user.click(screen.getByTestId('confirm-delete-btn'));
+		const confirmBtns = screen.getAllByRole('button', { name: /^Delete$/i });
+		await user.click(confirmBtns[confirmBtns.length - 1]);
 
-		await waitFor(
-			() => {
-				expect(deleteSpy).toHaveBeenCalled();
-				expect(screen.queryByDisplayValue('CI Bot')).not.toBeInTheDocument();
-			},
-			{ timeout: 3000 },
-		);
+		await waitFor(() => {
+			expect(deleteSpy).toHaveBeenCalled();
+		});
+
+		await waitFor(() => {
+			expect(screen.queryByDisplayValue('CI Bot')).not.toBeInTheDocument();
+		});
 	});
 
 	it('deleted account shows read-only name, no Save button, no Delete button', async () => {

frontend/src/constants/sessionStorage.ts

@@ -1,4 +1,3 @@
 export enum SESSIONSTORAGE {
 	RETRY_LAZY_REFRESHED = 'retry-lazy-refreshed',
-	VIEW_PANEL_HANDOFF = 'view-panel-handoff',
 }

frontend/src/container/LLMObservability/AttributeMapping/AttributeMappingsTab/AttributeMappingsTab.module.scss

@@ -1,7 +0,0 @@
-.pageError {
-	padding: var(--padding-3) var(--padding-4);
-	border-radius: var(--radius-2);
-	background: var(--callout-error-background);
-	color: var(--callout-error-title);
-	font-size: var(--periscope-font-size-base);
-}

frontend/src/container/LLMObservability/AttributeMapping/AttributeMappingsTab/AttributeMappingsTab.tsx

@@ -1,21 +0,0 @@
-import styles from './AttributeMappingsTab.module.scss';
-import MappingsTable from './components/MappingsTable/MappingsTable';
-import { useAttributeMappingStore } from './hooks/useAttributeMappingStore';
-
-function AttributeMappingsTab(): JSX.Element {
-	const store = useAttributeMappingStore();
-
-	return (
-		<div data-testid="attribute-mappings-tab">
-			{store.isError ? (
-				<div className={styles.pageError} role="alert">
-					Failed to load mapping groups. Please try again.
-				</div>
-			) : (
-				<MappingsTable store={store} />
-			)}
-		</div>
-	);
-}
-
-export default AttributeMappingsTab;

frontend/src/container/LLMObservability/AttributeMapping/AttributeMappingsTab/__tests__/AttributeMappingsTab.test.tsx

@@ -1,267 +0,0 @@
-import {
-	SpantypesFieldContextDTO as FieldContext,
-	SpantypesSpanMapperOperationDTO as MapperOperation,
-} from 'api/generated/services/sigNoz.schemas';
-import { rest, server } from 'mocks-server/server';
-import { render, screen, userEvent, waitFor, within } from 'tests/test-utils';
-
-import {
-	GROUPS_ENDPOINT,
-	makeGroupsResponse,
-	makeMapper,
-	makeMappersResponse,
-	mappersEndpoint,
-	mockGroups,
-	mockMappers,
-} from 'container/LLMObservability/AttributeMapping/__tests__/fixtures';
-import AttributeMappingsTab from '../AttributeMappingsTab';
-
-function setupGroups(groups = mockGroups): void {
-	server.use(
-		rest.get(GROUPS_ENDPOINT, (_req, res, ctx) =>
-			res(ctx.status(200), ctx.json(makeGroupsResponse(groups))),
-		),
-	);
-}
-
-function setupMappers(mappers = mockMappers, groupId = 'group-1'): void {
-	server.use(
-		rest.get(mappersEndpoint(groupId), (_req, res, ctx) =>
-			res(ctx.status(200), ctx.json(makeMappersResponse(mappers))),
-		),
-	);
-}
-
-async function expandGroup(
-	user: ReturnType<typeof userEvent.setup>,
-	groupId = 'group-1',
-): Promise<void> {
-	await user.click(screen.getByTestId(`group-expand-${groupId}`));
-}
-
-describe('AttributeMappingsTab (integration)', () => {
-	beforeEach(() => {
-		// Reset URL state between tests — jsdom shares window.location across a file.
-		window.history.pushState(null, '', '/');
-	});
-
-	afterEach(() => {
-		server.resetHandlers();
-	});
-
-	it('renders no error banner on a successful load', async () => {
-		setupGroups();
-		render(<AttributeMappingsTab />);
-
-		await waitFor(() =>
-			expect(screen.getByTestId('group-name-group-1')).toBeInTheDocument(),
-		);
-		expect(screen.queryByRole('alert')).not.toBeInTheDocument();
-	});
-
-	it('shows an error banner when the groups request fails', async () => {
-		server.use(
-			rest.get(GROUPS_ENDPOINT, (_req, res, ctx) => res(ctx.status(500))),
-		);
-		render(<AttributeMappingsTab />);
-
-		await expect(screen.findByRole('alert')).resolves.toHaveTextContent(
-			'Failed to load mapping groups. Please try again.',
-		);
-	});
-
-	it('shows the empty state when there are no groups', async () => {
-		setupGroups([]);
-		render(<AttributeMappingsTab />);
-
-		await expect(
-			screen.findByTestId('mapper-groups-empty'),
-		).resolves.toHaveTextContent('No mapping groups yet.');
-	});
-
-	it('renders each group header row with its name, condition count and status', async () => {
-		setupGroups();
-		render(<AttributeMappingsTab />);
-
-		// Condition filters are no longer shown inline as clauses — the header
-		// carries a count instead (the keys surface in the group drawer, later PR).
-		// Group headers are antd Collapse panels, so rows scope to the panel item.
-		// group-1: enabled, with attribute + resource condition keys.
-		const enabledRow = (await screen.findByTestId('group-name-group-1')).closest(
-			'.ant-collapse-item',
-		) as HTMLElement;
-		expect(
-			within(enabledRow).getByTestId('group-name-group-1'),
-		).toHaveTextContent('demo');
-		expect(
-			within(enabledRow).getByTestId('group-condition-count-group-1'),
-		).toHaveTextContent('2 conditions');
-		expect(within(enabledRow).getByTestId('group-enabled-group-1')).toBeChecked();
-
-		// group-2: disabled, with no condition keys.
-		const disabledRow = screen
-			.getByTestId('group-name-group-2')
-			.closest('.ant-collapse-item') as HTMLElement;
-		expect(within(disabledRow).getByText('Tool')).toBeInTheDocument();
-		expect(
-			within(disabledRow).getByTestId('group-condition-count-group-2'),
-		).toHaveTextContent('0 conditions');
-		expect(
-			within(disabledRow).getByTestId('group-enabled-group-2'),
-		).not.toBeChecked();
-	});
-
-	it('renders the group enable state as a read-only switch', async () => {
-		setupGroups();
-		render(<AttributeMappingsTab />);
-
-		// The status switch reflects enabled state but is non-interactive in this
-		// read-only listing — editing lands in a later PR.
-		const toggle = await screen.findByTestId('group-enabled-group-1');
-		expect(toggle).toBeChecked();
-		expect(toggle).toBeDisabled();
-	});
-
-	it("reveals a group's mappers on expand and hides them on collapse", async () => {
-		const user = userEvent.setup({ pointerEventsCheck: 0 });
-		setupGroups();
-		setupMappers([makeMapper({ id: 'mapper-1' })]);
-		render(<AttributeMappingsTab />);
-
-		await screen.findByTestId('group-name-group-1');
-		// The toggle is the antd Collapse header, which owns the expanded state.
-		const header = screen
-			.getByTestId('group-expand-group-1')
-			.closest('.ant-collapse-header') as HTMLElement;
-		expect(header).toHaveAttribute('aria-expanded', 'false');
-
-		await expandGroup(user);
-		expect(header).toHaveAttribute('aria-expanded', 'true');
-		await expect(
-			screen.findByTestId('mapper-target-mapper-1'),
-		).resolves.toBeInTheDocument();
-
-		await expandGroup(user);
-		await waitFor(() =>
-			expect(
-				screen.queryByTestId('mapper-target-mapper-1'),
-			).not.toBeInTheDocument(),
-		);
-	});
-
-	it("lazily fetches and renders a group's mappers on first expand", async () => {
-		const user = userEvent.setup({ pointerEventsCheck: 0 });
-		setupGroups();
-		setupMappers([
-			makeMapper({ id: 'mapper-1', name: 'gen_ai.request.model', enabled: true }),
-		]);
-		render(<AttributeMappingsTab />);
-
-		await screen.findByTestId('group-name-group-1');
-		// Mappers are not fetched until the row is expanded.
-		expect(
-			screen.queryByTestId('mapper-target-mapper-1'),
-		).not.toBeInTheDocument();
-
-		await expandGroup(user);
-
-		const target = await screen.findByTestId('mapper-target-mapper-1');
-		expect(target).toHaveTextContent('gen_ai.request.model');
-		const mapperRow = target.closest('tr') as HTMLElement;
-		// Sources ordered by priority, highest first (see fixtures).
-		const sources = within(mapperRow).getByTestId('mapper-sources-mapper-1');
-		expect(sources).toHaveTextContent('genai.model');
-		expect(sources).toHaveTextContent('llm.model');
-		// Writes-to field context + enabled status (an inline Switch, not text).
-		expect(within(mapperRow).getByText('attribute')).toBeInTheDocument();
-		expect(
-			within(mapperRow).getByTestId('mapper-enabled-mapper-1'),
-		).toBeChecked();
-	});
-
-	it("renders a mapper's enable state as a read-only switch", async () => {
-		const user = userEvent.setup({ pointerEventsCheck: 0 });
-		setupGroups();
-		setupMappers([makeMapper({ id: 'mapper-1', enabled: true })]);
-		render(<AttributeMappingsTab />);
-
-		await screen.findByTestId('group-name-group-1');
-		await expandGroup(user);
-
-		// Like the group switch, a mapper's status switch reflects state without
-		// accepting flips in this read-only listing.
-		const toggle = await screen.findByTestId('mapper-enabled-mapper-1');
-		expect(toggle).toBeChecked();
-		expect(toggle).toBeDisabled();
-	});
-
-	it('shows the mappers error state when the mappers request fails', async () => {
-		const user = userEvent.setup({ pointerEventsCheck: 0 });
-		setupGroups();
-		server.use(
-			rest.get(mappersEndpoint('group-1'), (_req, res, ctx) =>
-				res(ctx.status(500)),
-			),
-		);
-		render(<AttributeMappingsTab />);
-
-		await screen.findByTestId('group-name-group-1');
-		await expandGroup(user);
-
-		await expect(
-			screen.findByTestId('mappers-error-group-1'),
-		).resolves.toHaveTextContent('Failed to load mappings. Please try again.');
-	});
-
-	it('shows the mappers empty state when a group has no mappers', async () => {
-		const user = userEvent.setup({ pointerEventsCheck: 0 });
-		setupGroups();
-		setupMappers([]);
-		render(<AttributeMappingsTab />);
-
-		await screen.findByTestId('group-name-group-1');
-		await expandGroup(user);
-
-		await expect(
-			screen.findByTestId('mappers-empty-group-1'),
-		).resolves.toHaveTextContent('No mappings in this group yet.');
-	});
-
-	it('collapses extra mapper sources into a "+N more" label beyond the visible cap', async () => {
-		const user = userEvent.setup({ pointerEventsCheck: 0 });
-		setupGroups();
-		setupMappers([
-			makeMapper({
-				id: 'mapper-1',
-				config: {
-					sources: [1, 2, 3, 4, 5].map((priority) => ({
-						key: `source-${priority}`,
-						context: FieldContext.attribute,
-						operation: MapperOperation.copy,
-						priority,
-					})),
-				},
-			}),
-		]);
-		render(<AttributeMappingsTab />);
-
-		await screen.findByTestId('group-name-group-1');
-		await expandGroup(user);
-
-		await expect(screen.findByText('+2 more')).resolves.toBeInTheDocument();
-	});
-
-	it('shows a muted placeholder when a mapper has no sources', async () => {
-		const user = userEvent.setup({ pointerEventsCheck: 0 });
-		setupGroups();
-		setupMappers([makeMapper({ id: 'mapper-1', config: { sources: [] } })]);
-		render(<AttributeMappingsTab />);
-
-		await screen.findByTestId('group-name-group-1');
-		await expandGroup(user);
-
-		await waitFor(() =>
-			expect(screen.getByTestId('mapper-sources-mapper-1')).toHaveTextContent('—'),
-		);
-	});
-});

frontend/src/container/LLMObservability/AttributeMapping/AttributeMappingsTab/components/MappingsTable/GroupHeader/GroupHeader.module.scss

@@ -1,19 +0,0 @@
-.groupHeaderLabel {
-	display: flex;
-	align-items: center;
-	gap: var(--spacing-3);
-	min-width: 0;
-}
-
-.groupName {
-	color: var(--l1-foreground);
-	white-space: nowrap;
-	overflow: hidden;
-	text-overflow: ellipsis;
-}
-
-.groupCount {
-	color: var(--l3-foreground);
-	font-size: var(--font-size-xs);
-	white-space: nowrap;
-}

frontend/src/container/LLMObservability/AttributeMapping/AttributeMappingsTab/components/MappingsTable/GroupHeader/GroupHeader.tsx

@@ -1,36 +0,0 @@
-import { Typography } from '@signozhq/ui/typography';
-import { MappingGroup } from 'container/LLMObservability/AttributeMapping/types';
-
-import styles from './GroupHeader.module.scss';
-
-interface GroupHeaderProps {
-	group: MappingGroup;
-}
-
-function GroupHeader({ group }: GroupHeaderProps): JSX.Element {
-	const conditionCount = group.attributes.length + group.resource.length;
-
-	return (
-		<div
-			className={styles.groupHeaderLabel}
-			data-testid={`group-expand-${group.id}`}
-		>
-			<Typography.Text
-				as="span"
-				className={styles.groupName}
-				testId={`group-name-${group.id}`}
-			>
-				{group.name}
-			</Typography.Text>
-			<Typography.Text
-				as="span"
-				className={styles.groupCount}
-				testId={`group-condition-count-${group.id}`}
-			>
-				· {conditionCount} {conditionCount === 1 ? 'condition' : 'conditions'}
-			</Typography.Text>
-		</div>
-	);
-}
-
-export default GroupHeader;

frontend/src/container/LLMObservability/AttributeMapping/AttributeMappingsTab/components/MappingsTable/GroupHeader/index.ts

@@ -1 +0,0 @@
-export { default } from './GroupHeader';

frontend/src/container/LLMObservability/AttributeMapping/AttributeMappingsTab/components/MappingsTable/GroupHeaderActions/GroupHeaderActions.module.scss

@@ -1,6 +0,0 @@
-.actions {
-	display: flex;
-	align-items: center;
-	justify-content: flex-end;
-	gap: var(--spacing-3);
-}

frontend/src/container/LLMObservability/AttributeMapping/AttributeMappingsTab/components/MappingsTable/GroupHeaderActions/GroupHeaderActions.tsx

@@ -1,26 +0,0 @@
-import { Switch } from '@signozhq/ui/switch';
-
-import { MappingGroup } from 'container/LLMObservability/AttributeMapping/types';
-import styles from './GroupHeaderActions.module.scss';
-
-interface GroupHeaderActionsProps {
-	group: MappingGroup;
-}
-
-function GroupHeaderActions({ group }: GroupHeaderActionsProps): JSX.Element {
-	return (
-		<div
-			className={styles.actions}
-			onClick={(event): void => event.stopPropagation()}
-		>
-			<Switch
-				value={group.enabled}
-				// We don't yet support toggling a group's enabled state in this read-only PR, so disable the switch. A later PR will add the toggle handler and its drawer.
-				disabled
-				testId={`group-enabled-${group.id}`}
-			/>
-		</div>
-	);
-}
-
-export default GroupHeaderActions;

frontend/src/container/LLMObservability/AttributeMapping/AttributeMappingsTab/components/MappingsTable/GroupHeaderActions/index.ts

@@ -1 +0,0 @@
-export { default } from './GroupHeaderActions';

frontend/src/container/LLMObservability/AttributeMapping/AttributeMappingsTab/components/MappingsTable/GroupMappers/GroupMappers.module.scss

@@ -1,14 +0,0 @@
-.table {
-	width: 100%;
-	border-collapse: collapse;
-	table-layout: fixed;
-}
-
-.mapperStateRow .stateCell {
-	color: var(--l3-foreground);
-}
-
-.stateCell {
-	padding: var(--spacing-4) var(--spacing-6) var(--spacing-4) var(--spacing-12);
-	font-size: var(--periscope-font-size-base);
-}

frontend/src/container/LLMObservability/AttributeMapping/AttributeMappingsTab/components/MappingsTable/GroupMappers/GroupMappers.tsx

@@ -1,98 +0,0 @@
-import { useListSpanMappers } from 'api/generated/services/spanmapper';
-import { motion } from 'motion/react';
-
-import {
-	MappingGroup,
-	Mapping,
-} from 'container/LLMObservability/AttributeMapping/types';
-import { buildMappingsFromListResponse } from 'container/LLMObservability/AttributeMapping/utils';
-import { COLUMN_COUNT } from '../constants';
-import MapperRow, { MapperRowSkeleton } from '../MapperRow';
-import MappingsColgroup from '../MappingsColgroup';
-import styles from './GroupMappers.module.scss';
-
-const MAPPER_SKELETON_ROWS = 1;
-
-const STATE_ROW_MOTION = {
-	initial: { opacity: 0 },
-	animate: { opacity: 1 },
-	transition: { duration: 0.18, ease: 'easeOut' },
-} as const;
-
-interface StateRowProps {
-	groupId: string;
-}
-
-function ErrorRow({ groupId }: StateRowProps): JSX.Element {
-	return (
-		<motion.tr className={styles.mapperStateRow} {...STATE_ROW_MOTION}>
-			<td
-				colSpan={COLUMN_COUNT}
-				className={styles.stateCell}
-				data-testid={`mappers-error-${groupId}`}
-			>
-				Failed to load mappings. Please try again.
-			</td>
-		</motion.tr>
-	);
-}
-
-function EmptyRow({ groupId }: StateRowProps): JSX.Element {
-	return (
-		<motion.tr className={styles.mapperStateRow} {...STATE_ROW_MOTION}>
-			<td
-				colSpan={COLUMN_COUNT}
-				className={styles.stateCell}
-				data-testid={`mappers-empty-${groupId}`}
-			>
-				No mappings in this group yet.
-			</td>
-		</motion.tr>
-	);
-}
-
-interface GroupMappersProps {
-	group: MappingGroup;
-}
-
-function GroupMappers({ group }: GroupMappersProps): JSX.Element {
-	const {
-		data: mappers = [],
-		isLoading,
-		isError,
-	} = useListSpanMappers<Mapping[]>(
-		{
-			groupId: group.id,
-		},
-		{
-			query: {
-				refetchOnMount: false,
-				select: buildMappingsFromListResponse,
-			},
-		},
-	);
-
-	let rows: JSX.Element[];
-	if (isError) {
-		rows = [<ErrorRow key="error" groupId={group.id} />];
-	} else if (isLoading) {
-		rows = Array.from({ length: MAPPER_SKELETON_ROWS }).map((_, index) => (
-			<MapperRowSkeleton key={`mapper-skeleton-${index}`} />
-		));
-	} else if (mappers.length === 0) {
-		rows = [<EmptyRow key="empty" groupId={group.id} />];
-	} else {
-		rows = mappers.map((mapper, index) => (
-			<MapperRow key={mapper.id} mapper={mapper} index={index} />
-		));
-	}
-
-	return (
-		<table className={styles.table}>
-			<MappingsColgroup />
-			<tbody>{rows}</tbody>
-		</table>
-	);
-}
-
-export default GroupMappers;

frontend/src/container/LLMObservability/AttributeMapping/AttributeMappingsTab/components/MappingsTable/GroupMappers/index.ts

@@ -1 +0,0 @@
-export { default } from './GroupMappers';

frontend/src/container/LLMObservability/AttributeMapping/AttributeMappingsTab/components/MappingsTable/MapperRow/MapperRow.module.scss

@@ -1,65 +0,0 @@
-.mapperRow {
-	&:hover {
-		background: var(--l2-background-hover);
-	}
-}
-
-.cell {
-	padding: var(--spacing-4) var(--spacing-6);
-	vertical-align: middle;
-	color: var(--l1-foreground);
-	font-size: var(--periscope-font-size-base);
-}
-
-// Indent the first cell so mapper rows read as nested under their group.
-.targetCell {
-	padding-left: var(--spacing-12);
-}
-
-// Shorter vertical padding so the loading state reads as a compact placeholder.
-.skeletonCell {
-	composes: cell;
-	:global(.ant-skeleton-input) {
-		min-height: 18px !important;
-		height: 18px !important;
-	}
-	:global(.ant-skeleton-button) {
-		min-height: 18px !important;
-		height: 18px !important;
-	}
-}
-
-.statusCell {
-	text-align: right;
-}
-
-.rowActions {
-	display: flex;
-	align-items: center;
-	justify-content: flex-end;
-	gap: var(--spacing-3);
-}
-
-.sources {
-	display: flex;
-	flex-wrap: wrap;
-	align-items: center;
-	gap: var(--spacing-3);
-}
-
-.sourceChipText {
-	display: block;
-	max-width: 220px;
-	overflow: hidden;
-	text-overflow: ellipsis;
-	white-space: nowrap;
-}
-
-.sourceMore {
-	font-size: var(--font-size-xs);
-	white-space: nowrap;
-}
-
-.muted {
-	color: var(--l3-foreground);
-}

frontend/src/container/LLMObservability/AttributeMapping/AttributeMappingsTab/components/MappingsTable/MapperRow/MapperRow.tsx

@@ -1,102 +0,0 @@
-import { Badge } from '@signozhq/ui/badge';
-import { Switch } from '@signozhq/ui/switch';
-import { Typography } from '@signozhq/ui/typography';
-import { SpantypesFieldContextDTO } from 'api/generated/services/sigNoz.schemas';
-import cx from 'classnames';
-import { motion } from 'motion/react';
-
-import { Mapping } from 'container/LLMObservability/AttributeMapping/types';
-import styles from './MapperRow.module.scss';
-
-const MAX_VISIBLE_SOURCES = 3;
-
-const ROW_TRANSITION = { duration: 0.18, ease: 'easeOut' } as const;
-const MAX_STAGGERED_ROWS = 6;
-const STAGGER_STEP = 0.03;
-
-interface MapperRowProps {
-	mapper: Mapping;
-	index: number;
-}
-
-function MapperRow({ mapper, index }: MapperRowProps): JSX.Element {
-	const sources = mapper.sources ?? [];
-	const visibleSources = sources.slice(0, MAX_VISIBLE_SOURCES);
-	const remainingSources = sources.length - visibleSources.length;
-
-	return (
-		<motion.tr
-			className={styles.mapperRow}
-			data-testid={`mapper-row-${mapper.id}`}
-			initial={{ opacity: 0, y: -4 }}
-			animate={{ opacity: 1, y: 0 }}
-			transition={{
-				...ROW_TRANSITION,
-				delay: Math.min(index, MAX_STAGGERED_ROWS) * STAGGER_STEP,
-			}}
-		>
-			<td className={cx(styles.cell, styles.targetCell)}>
-				<Typography.Text
-					truncate={1}
-					title={mapper.name}
-					data-testid={`mapper-target-${mapper.id}`}
-				>
-					{mapper.name}
-				</Typography.Text>
-			</td>
-			<td className={styles.cell}>
-				{sources.length === 0 ? (
-					<span className={styles.muted} data-testid={`mapper-sources-${mapper.id}`}>
-						—
-					</span>
-				) : (
-					<div
-						className={styles.sources}
-						data-testid={`mapper-sources-${mapper.id}`}
-					>
-						{visibleSources.map((source) => (
-							<Badge
-								variant="outline"
-								color="vanilla"
-								className={styles.sourceChip}
-								key={`${source.context}:${source.key}`}
-							>
-								<span className={styles.sourceChipText} title={source.key}>
-									{source.key}
-								</span>
-							</Badge>
-						))}
-						{remainingSources > 0 && (
-							<span className={cx(styles.sourceMore, styles.muted)}>
-								+{remainingSources} more
-							</span>
-						)}
-					</div>
-				)}
-			</td>
-			<td className={styles.cell}>
-				<Badge
-					color={
-						mapper.fieldContext === SpantypesFieldContextDTO.resource
-							? 'amber'
-							: 'robin'
-					}
-					variant="outline"
-				>
-					{mapper.fieldContext}
-				</Badge>
-			</td>
-			<td className={cx(styles.cell, styles.statusCell)}>
-				<div className={styles.rowActions}>
-					<Switch
-						value={mapper.enabled}
-						disabled
-						testId={`mapper-enabled-${mapper.id}`}
-					/>
-				</div>
-			</td>
-		</motion.tr>
-	);
-}
-
-export default MapperRow;

frontend/src/container/LLMObservability/AttributeMapping/AttributeMappingsTab/components/MappingsTable/MapperRow/MapperRowSkeleton.tsx

@@ -1,30 +0,0 @@
-import { Skeleton } from 'antd';
-import cx from 'classnames';
-
-import styles from './MapperRow.module.scss';
-
-function MapperRowSkeleton(): JSX.Element {
-	return (
-		<tr className={styles.mapperRow}>
-			<td className={cx(styles.skeletonCell, styles.targetCell)}>
-				<Skeleton.Input active size="small" style={{ width: '55%' }} />
-			</td>
-			<td className={styles.skeletonCell}>
-				<div className={styles.sources}>
-					<Skeleton.Button active size="small" style={{ width: 88 }} />
-					<Skeleton.Button active size="small" style={{ width: 56 }} />
-				</div>
-			</td>
-			<td className={styles.skeletonCell}>
-				<Skeleton.Button active size="small" style={{ width: 72 }} />
-			</td>
-			<td className={cx(styles.skeletonCell, styles.statusCell)}>
-				<div className={styles.rowActions}>
-					<Skeleton.Button active size="small" shape="round" />
-				</div>
-			</td>
-		</tr>
-	);
-}
-
-export default MapperRowSkeleton;

frontend/src/container/LLMObservability/AttributeMapping/AttributeMappingsTab/components/MappingsTable/MapperRow/index.ts

@@ -1,2 +0,0 @@
-export { default } from './MapperRow';
-export { default as MapperRowSkeleton } from './MapperRowSkeleton';

frontend/src/container/LLMObservability/AttributeMapping/AttributeMappingsTab/components/MappingsTable/MappingsColgroup/MappingsColgroup.module.scss

@@ -1,11 +0,0 @@
-.colTarget {
-	width: 32%;
-}
-
-.colWritesTo {
-	width: 140px;
-}
-
-.colStatus {
-	width: 120px;
-}

frontend/src/container/LLMObservability/AttributeMapping/AttributeMappingsTab/components/MappingsTable/MappingsColgroup/MappingsColgroup.tsx

@@ -1,14 +0,0 @@
-import styles from './MappingsColgroup.module.scss';
-
-function MappingsColgroup(): JSX.Element {
-	return (
-		<colgroup>
-			<col className={styles.colTarget} />
-			<col />
-			<col className={styles.colWritesTo} />
-			<col className={styles.colStatus} />
-		</colgroup>
-	);
-}
-
-export default MappingsColgroup;

frontend/src/container/LLMObservability/AttributeMapping/AttributeMappingsTab/components/MappingsTable/MappingsColgroup/index.ts

@@ -1 +0,0 @@
-export { default } from './MappingsColgroup';

frontend/src/container/LLMObservability/AttributeMapping/AttributeMappingsTab/components/MappingsTable/MappingsTable.module.scss

@@ -1,124 +0,0 @@
-.table {
-	width: 100%;
-	border-collapse: collapse;
-	table-layout: fixed;
-}
-
-.headerRow {
-	border-bottom: 1px solid var(--l2-border);
-}
-
-.headerCell {
-	padding: var(--spacing-4) var(--spacing-6);
-	text-align: left;
-	font-size: var(--periscope-font-size-base);
-	font-weight: var(--font-weight-normal);
-	color: var(--l2-foreground);
-
-	&:last-child {
-		text-align: right;
-	}
-}
-
-.groupsCollapse:global(.ant-collapse) {
-	background: transparent;
-	border: none;
-	border-radius: 0;
-
-	> :global(.ant-collapse-item) {
-		border-bottom: none;
-		border-top: 1px solid var(--l2-border);
-		border-radius: 0;
-
-		&:last-child {
-			border-bottom: 1px solid var(--l2-border);
-			border-radius: 0;
-		}
-
-		> :global(.ant-collapse-header) {
-			align-items: center;
-			gap: var(--spacing-3);
-			background: var(--l2-background);
-			border-radius: 0;
-			padding: var(--spacing-3) var(--spacing-6);
-			color: var(--l3-foreground);
-
-			:global(.ant-collapse-expand-icon) {
-				display: flex;
-				align-items: center;
-				height: auto;
-				padding-inline-end: 0;
-				color: var(--l3-foreground);
-			}
-
-			:global(.ant-collapse-header-text) {
-				min-width: 0;
-			}
-
-			:global(.ant-collapse-extra) {
-				display: flex;
-				align-items: center;
-			}
-		}
-	}
-
-	:global(.ant-collapse-content) {
-		background: transparent;
-		border-top: none;
-		color: inherit;
-
-		> :global(.ant-collapse-content-box) {
-			padding: 0;
-		}
-	}
-}
-
-.tableEmpty {
-	padding: var(--spacing-12) var(--spacing-6);
-	text-align: center;
-	color: var(--l3-foreground);
-	font-size: var(--periscope-font-size-base);
-}
-
-.skeletonList {
-	display: flex;
-	flex-direction: column;
-}
-
-// Mirrors the Collapse header banner while groups load.
-.skeletonBanner {
-	display: flex;
-	align-items: center;
-	justify-content: space-between;
-	gap: var(--spacing-4);
-	padding: var(--spacing-3) var(--spacing-6);
-	background: var(--l2-background);
-	border-top: 1px solid var(--l2-border);
-
-	&:last-child {
-		border-bottom: 1px solid var(--l2-border);
-	}
-	:global(.ant-skeleton-input) {
-		min-height: 18px !important;
-		height: 18px !important;
-	}
-	:global(.ant-skeleton-button) {
-		min-height: 18px !important;
-		height: 18px !important;
-	}
-}
-
-.skeletonGroupLeft {
-	display: flex;
-	align-items: center;
-	gap: var(--spacing-3);
-	flex: 1;
-	min-width: 0;
-}
-
-.skeletonGroupRight {
-	display: flex;
-	align-items: center;
-	gap: var(--spacing-3);
-	flex-shrink: 0;
-}

frontend/src/container/LLMObservability/AttributeMapping/AttributeMappingsTab/components/MappingsTable/MappingsTable.tsx

@@ -1,96 +0,0 @@
-import { useState } from 'react';
-import { ChevronDown, ChevronRight } from '@signozhq/icons';
-import { Collapse, type CollapseProps, Skeleton } from 'antd';
-
-import { AttributeMappingStore } from 'container/LLMObservability/AttributeMapping/AttributeMappingsTab/hooks/useAttributeMappingStore';
-import GroupHeader from './GroupHeader';
-import GroupHeaderActions from './GroupHeaderActions';
-import GroupMappers from './GroupMappers';
-import MappingsColgroup from './MappingsColgroup';
-import styles from './MappingsTable.module.scss';
-
-const SKELETON_ROW_COUNT = 3;
-
-interface MappingsTableProps {
-	store: AttributeMappingStore;
-}
-
-function MappingsTable({ store }: MappingsTableProps): JSX.Element {
-	const [expandedGroups, setExpandedGroups] = useState<string[]>([]);
-
-	const isEmpty = !store.isLoading && store.groups.length === 0;
-
-	const items: CollapseProps['items'] = store.groups.map((group) => ({
-		key: group.id,
-		label: <GroupHeader group={group} />,
-		extra: <GroupHeaderActions group={group} />,
-		children: <GroupMappers group={group} />,
-	}));
-
-	const skeletonBanners = (
-		<div className={styles.skeletonList}>
-			{Array.from({ length: SKELETON_ROW_COUNT }).map((_, index) => (
-				<div
-					// eslint-disable-next-line react/no-array-index-key
-					key={`group-skeleton-${index}`}
-					className={styles.skeletonBanner}
-				>
-					<div className={styles.skeletonGroupLeft}>
-						<Skeleton.Input
-							active
-							size="small"
-							style={{ width: index % 2 === 0 ? 200 : 140 }}
-						/>
-						<Skeleton.Input active size="small" style={{ width: 64 }} />
-					</div>
-					<div className={styles.skeletonGroupRight}>
-						<Skeleton.Button active size="small" shape="round" />
-					</div>
-				</div>
-			))}
-		</div>
-	);
-
-	if (isEmpty) {
-		return (
-			<div className={styles.tableEmpty} data-testid="mapper-groups-empty">
-				No mapping groups yet.
-			</div>
-		);
-	}
-
-	return (
-		<div data-testid="mappings-table">
-			<table className={styles.table}>
-				<MappingsColgroup />
-				<thead>
-					<tr className={styles.headerRow}>
-						<th className={styles.headerCell}>Target</th>
-						<th className={styles.headerCell}>Sources</th>
-						<th className={styles.headerCell}>Writes to</th>
-						<th className={styles.headerCell}>Status</th>
-					</tr>
-				</thead>
-			</table>
-			{store.isLoading ? (
-				skeletonBanners
-			) : (
-				<Collapse
-					className={styles.groupsCollapse}
-					activeKey={expandedGroups}
-					onChange={(keys): void =>
-						setExpandedGroups(Array.isArray(keys) ? keys : [keys])
-					}
-					bordered={false}
-					destroyInactivePanel
-					expandIcon={({ isActive }): JSX.Element =>
-						isActive ? <ChevronDown size={14} /> : <ChevronRight size={14} />
-					}
-					items={items}
-				/>
-			)}
-		</div>
-	);
-}
-
-export default MappingsTable;

frontend/src/container/LLMObservability/AttributeMapping/AttributeMappingsTab/components/MappingsTable/constants.ts

@@ -1 +0,0 @@
-export const COLUMN_COUNT = 4;

frontend/src/container/LLMObservability/AttributeMapping/AttributeMappingsTab/hooks/useAttributeMappingStore.ts

@@ -1,33 +0,0 @@
-import { useMemo } from 'react';
-import { SpantypesSpanMapperGroupDTO } from 'api/generated/services/sigNoz.schemas';
-import { useListSpanMapperGroups } from 'api/generated/services/spanmapper';
-
-import { MappingGroup } from 'container/LLMObservability/AttributeMapping/types';
-import { buildMappingGroup } from 'container/LLMObservability/AttributeMapping/utils';
-
-export interface AttributeMappingStore {
-	groups: MappingGroup[];
-	isLoading: boolean;
-	isError: boolean;
-}
-
-// Read-only store for the listing view: loads the server groups only. Each
-// group's mappers are fetched lazily when its panel is expanded (see
-// GroupMappers), so page load is a single request instead of an N+1 fan-out
-// across every group. Editing (enabled toggles, save/discard) and its drawers
-// land in a later PR — this PR only lists.
-export function useAttributeMappingStore(): AttributeMappingStore {
-	const groupsQuery = useListSpanMapperGroups();
-
-	const groups = useMemo<MappingGroup[]>(() => {
-		const serverGroups: SpantypesSpanMapperGroupDTO[] =
-			groupsQuery.data?.data?.items ?? [];
-		return serverGroups.map((group) => buildMappingGroup(group));
-	}, [groupsQuery.data]);
-
-	return {
-		groups,
-		isLoading: groupsQuery.isLoading,
-		isError: groupsQuery.isError,
-	};
-}

frontend/src/container/LLMObservability/AttributeMapping/LLMObservabilityAttributeMapping.module.scss

@@ -2,5 +2,12 @@
 	display: flex;
 	flex-direction: column;
 	gap: var(--spacing-8);
-	padding: var(--spacing-0);
+	padding: var(--spacing-12);
+}
+
+.tableEmpty {
+	padding: var(--spacing-12) var(--spacing-6);
+	text-align: center;
+	color: var(--l3-foreground);
+	font-size: var(--periscope-font-size-base);
 }

frontend/src/container/LLMObservability/AttributeMapping/LLMObservabilityAttributeMapping.tsx

@@ -1,27 +1,9 @@
-import { Tabs } from '@signozhq/ui/tabs';
-
 import AttributeMappingHeader from './components/AttributeMappingHeader';
-import AttributeMappingsTab from './AttributeMappingsTab/AttributeMappingsTab';
 import styles from './LLMObservabilityAttributeMapping.module.scss';
 
 const noop = (): void => undefined;
 
 function LLMObservabilityAttributeMapping(): JSX.Element {
-	const tabItems = [
-		{
-			key: 'attribute-mappings',
-			label: 'Attribute mappings',
-			children: <AttributeMappingsTab />,
-		},
-		{
-			key: 'test',
-			label: 'Test',
-			disabled: true,
-			disabledReason: 'Coming soon',
-			children: null,
-		},
-	];
-
 	return (
 		<div
 			className={styles.llmObservabilityAttributeMapping}
@@ -34,11 +16,9 @@ function LLMObservabilityAttributeMapping(): JSX.Element {
 				onSave={noop}
 			/>
 
-			<Tabs
-				testId="attribute-mapping-tabs"
-				defaultValue="attribute-mappings"
-				items={tabItems}
-			/>
+			<div className={styles.tableEmpty} data-testid="attribute-mapping-empty">
+				No mapping groups configured yet.
+			</div>
 		</div>
 	);
 }

frontend/src/container/LLMObservability/AttributeMapping/__tests__/LLMObservabilityAttributeMapping.test.tsx

@@ -1,67 +0,0 @@
-import { rest, server } from 'mocks-server/server';
-import { render, screen } from 'tests/test-utils';
-
-import LLMObservabilityAttributeMapping from '../LLMObservabilityAttributeMapping';
-import { GROUPS_ENDPOINT, makeGroupsResponse, mockGroups } from './fixtures';
-
-function setupGroups(): void {
-	server.use(
-		rest.get(GROUPS_ENDPOINT, (_req, res, ctx) =>
-			res(ctx.status(200), ctx.json(makeGroupsResponse(mockGroups))),
-		),
-	);
-}
-
-describe('LLMObservabilityAttributeMapping', () => {
-	beforeEach(() => {
-		window.history.pushState(null, '', '/');
-		setupGroups();
-	});
-
-	afterEach(() => {
-		server.resetHandlers();
-	});
-
-	it('renders the page shell', () => {
-		render(<LLMObservabilityAttributeMapping />);
-
-		expect(
-			screen.getByTestId('llm-observability-attribute-mapping-page'),
-		).toBeInTheDocument();
-	});
-
-	it('shows the attribute-mappings and test sub-tab labels', () => {
-		render(<LLMObservabilityAttributeMapping />);
-
-		expect(
-			screen.getByRole('tab', { name: 'Attribute mappings' }),
-		).toBeInTheDocument();
-		expect(screen.getByRole('tab', { name: 'Test' })).toBeInTheDocument();
-	});
-
-	it('activates the attribute-mappings tab by default and renders its content', async () => {
-		render(<LLMObservabilityAttributeMapping />);
-
-		const attributeMappingsTab = screen.getByRole('tab', {
-			name: 'Attribute mappings',
-		});
-		expect(attributeMappingsTab).toHaveAttribute('data-state', 'active');
-		await expect(
-			screen.findByTestId('attribute-mappings-tab'),
-		).resolves.toBeInTheDocument();
-	});
-
-	it('renders the header with its description and no Save/Discard while pristine', () => {
-		render(<LLMObservabilityAttributeMapping />);
-
-		expect(
-			screen.getByText(
-				'Configure source-to-target attribute remapping for LLM traces',
-			),
-		).toBeInTheDocument();
-		// The actions only appear once there are staged changes.
-		expect(screen.queryByTestId('save-changes-btn')).not.toBeInTheDocument();
-		expect(screen.queryByTestId('discard-changes-btn')).not.toBeInTheDocument();
-		expect(screen.queryByTestId('unsaved-changes')).not.toBeInTheDocument();
-	});
-});

frontend/src/container/LLMObservability/AttributeMapping/__tests__/fixtures.ts

@@ -1,93 +0,0 @@
-import {
-	SpantypesFieldContextDTO as FieldContext,
-	SpantypesSpanMapperDTO as Mapper,
-	SpantypesSpanMapperGroupDTO as MapperGroup,
-	SpantypesSpanMapperOperationDTO as MapperOperation,
-} from 'api/generated/services/sigNoz.schemas';
-
-// Endpoint globs used by MSW handlers. The generated client hits relative
-// `/api/v1/span_mapper_groups[...]`, so the `*` prefix matches regardless of
-// base URL.
-export const GROUPS_ENDPOINT = '*/api/v1/span_mapper_groups';
-export function mappersEndpoint(groupId: string): string {
-	return `*/api/v1/span_mapper_groups/${groupId}/span_mappers`;
-}
-
-export function makeGroup(overrides: Partial<MapperGroup> = {}): MapperGroup {
-	return {
-		id: 'group-1',
-		orgId: 'org-1',
-		name: 'demo',
-		enabled: true,
-		condition: {
-			attributes: ['ai.embeddings'],
-			resource: ['cloud.account.id'],
-		},
-		...overrides,
-	};
-}
-
-export function makeMapper(overrides: Partial<Mapper> = {}): Mapper {
-	return {
-		id: 'mapper-1',
-		group_id: 'group-1',
-		name: 'gen_ai.request.model',
-		enabled: true,
-		fieldContext: FieldContext.attribute,
-		config: {
-			sources: [
-				{
-					key: 'genai.model',
-					context: FieldContext.attribute,
-					operation: MapperOperation.copy,
-					priority: 2,
-				},
-				{
-					key: 'llm.model',
-					context: FieldContext.attribute,
-					operation: MapperOperation.move,
-					priority: 1,
-				},
-			],
-		},
-		...overrides,
-	};
-}
-
-// Both list endpoints share the same `{ status, data: { items } }` envelope —
-// the generated schema mis-types the mappers response with the groups DTO
-// (see GroupMappers), but the runtime envelope shape is identical.
-export function makeGroupsResponse(groups: MapperGroup[]): {
-	status: string;
-	data: { items: MapperGroup[] };
-} {
-	return { status: 'ok', data: { items: groups } };
-}
-
-export function makeMappersResponse(mappers: Mapper[]): {
-	status: string;
-	data: { items: Mapper[] };
-} {
-	return { status: 'ok', data: { items: mappers } };
-}
-
-export const mockGroups: MapperGroup[] = [
-	makeGroup({
-		id: 'group-1',
-		name: 'demo',
-		condition: {
-			attributes: ['ai.embeddings'],
-			resource: ['cloud.account.id'],
-		},
-	}),
-	makeGroup({
-		id: 'group-2',
-		name: 'Tool',
-		enabled: false,
-		condition: { attributes: null, resource: null },
-	}),
-];
-
-export const mockMappers: Mapper[] = [
-	makeMapper({ id: 'mapper-1', group_id: 'group-1' }),
-];

frontend/src/container/LLMObservability/AttributeMapping/components/AttributeMappingHeader/AttributeMappingHeader.module.scss

@@ -5,6 +5,23 @@
 	gap: var(--spacing-8);
 }
 
+.pageHeaderTitle {
+	display: flex;
+	flex-direction: column;
+}
+
+.title {
+	margin: 0;
+	font-size: var(--periscope-font-size-large);
+	font-weight: var(--font-weight-semibold);
+}
+
+.description {
+	margin: var(--spacing-2) 0 0;
+	font-size: var(--periscope-font-size-base);
+	color: var(--l3-foreground);
+}
+
 .pageHeaderActions {
 	display: flex;
 	align-items: center;

frontend/src/container/LLMObservability/AttributeMapping/components/AttributeMappingHeader/AttributeMappingHeader.tsx

@@ -1,5 +1,4 @@
 import { Button } from '@signozhq/ui/button';
-import { Typography } from '@signozhq/ui/typography';
 
 import styles from './AttributeMappingHeader.module.scss';
 
@@ -18,35 +17,38 @@ function AttributeMappingHeader({
 }: AttributeMappingHeaderProps): JSX.Element {
 	return (
 		<header className={styles.pageHeader}>
-			<Typography.Text as="p" size="base" color="muted">
-				Configure source-to-target attribute remapping for LLM traces
-			</Typography.Text>
-			{isDirty && (
-				<div className={styles.pageHeaderActions}>
+			<div className={styles.pageHeaderTitle}>
+				<h1 className={styles.title}>Attribute Mapping</h1>
+				<p className={styles.description}>
+					Configure source-to-target attribute remapping for LLM traces
+				</p>
+			</div>
+			<div className={styles.pageHeaderActions}>
+				{isDirty && (
 					<span className={styles.unsavedChanges} data-testid="unsaved-changes">
 						Unsaved changes
 					</span>
-					<Button
-						variant="outlined"
-						color="secondary"
-						onClick={onDiscard}
-						disabled={isSaving}
-						testId="discard-changes-btn"
-					>
-						Discard
-					</Button>
-					<Button
-						variant="solid"
-						color="primary"
-						onClick={onSave}
-						loading={isSaving}
-						disabled={isSaving}
-						testId="save-changes-btn"
-					>
-						{isSaving ? 'Saving…' : 'Save changes'}
-					</Button>
-				</div>
-			)}
+				)}
+				<Button
+					variant="outlined"
+					color="secondary"
+					onClick={onDiscard}
+					disabled={!isDirty || isSaving}
+					testId="discard-changes-btn"
+				>
+					Discard
+				</Button>
+				<Button
+					variant="solid"
+					color="primary"
+					onClick={onSave}
+					loading={isSaving}
+					disabled={!isDirty || isSaving}
+					testId="save-changes-btn"
+				>
+					{isSaving ? 'Saving…' : 'Save changes'}
+				</Button>
+			</div>
 		</header>
 	);
 }

frontend/src/container/LLMObservability/AttributeMapping/types.ts

@@ -1,26 +0,0 @@
-import {
-	SpantypesFieldContextDTO,
-	SpantypesSpanMapperOperationDTO,
-} from 'api/generated/services/sigNoz.schemas';
-
-export interface SourceConfig {
-	key: string;
-	context: SpantypesFieldContextDTO;
-	operation: SpantypesSpanMapperOperationDTO;
-}
-
-export interface Mapping {
-	id: string;
-	name: string;
-	fieldContext: SpantypesFieldContextDTO;
-	sources: SourceConfig[];
-	enabled: boolean;
-}
-
-export interface MappingGroup {
-	id: string;
-	name: string;
-	attributes: string[];
-	resource: string[];
-	enabled: boolean;
-}

frontend/src/container/LLMObservability/AttributeMapping/utils.ts

@@ -1,48 +0,0 @@
-import {
-	ListSpanMappers200,
-	SpantypesSpanMapperDTO,
-	SpantypesSpanMapperGroupDTO,
-} from 'api/generated/services/sigNoz.schemas';
-
-import { MappingGroup, Mapping, SourceConfig } from './types';
-
-function getMapperSources(mapper: SpantypesSpanMapperDTO): SourceConfig[] {
-	const sources = mapper.config?.sources ?? [];
-	return [...sources]
-		.sort((a, b) => a.priority - b.priority)
-		.map((source) => ({
-			key: source.key,
-			context: source.context,
-			operation: source.operation,
-		}));
-}
-
-export function buildMapping(mapper: SpantypesSpanMapperDTO): Mapping {
-	return {
-		id: mapper.id,
-		name: mapper.name,
-		fieldContext: mapper.fieldContext,
-		sources: getMapperSources(mapper),
-		enabled: mapper.enabled,
-	};
-}
-
-export function buildMappingsFromListResponse(
-	response: ListSpanMappers200,
-): Mapping[] {
-	const items = (response.data?.items ??
-		[]) as unknown as SpantypesSpanMapperDTO[];
-	return items.map(buildMapping);
-}
-
-export function buildMappingGroup(
-	group: SpantypesSpanMapperGroupDTO,
-): MappingGroup {
-	return {
-		id: group.id,
-		name: group.name,
-		attributes: group.condition?.attributes ?? [],
-		resource: group.condition?.resource ?? [],
-		enabled: group.enabled,
-	};
-}

frontend/src/container/ServiceAccountsSettings/__tests__/ServiceAccountsSettings.authz.integration.test.tsx

@@ -208,7 +208,7 @@ describe('ServiceAccountsSettings (integration)', () => {
 
 		fireEvent.click(screen.getByRole('button', { name: /New Service Account/i }));
 
-		await screen.findByTestId('create-service-account-modal');
+		await screen.findByRole('dialog', { name: /New Service Account/i });
 		expect(screen.getByPlaceholderText('Enter a name')).toBeInTheDocument();
 	});
 

frontend/src/hooks/useExportData/__tests__/useClientExport.test.ts

@@ -1,130 +0,0 @@
-import { act, renderHook } from '@testing-library/react';
-import {
-	downloadFile,
-	getTimestampedFileName,
-} from 'lib/exportData/downloadFile';
-import { ExportFormat } from 'lib/exportData/types';
-import { QueryRangeResponseV5 } from 'types/api/v5/queryRange';
-
-import { useClientExport } from '../useClientExport';
-
-jest.mock('lib/exportData/downloadFile', () => ({
-	...jest.requireActual('lib/exportData/downloadFile'),
-	downloadFile: jest.fn(),
-}));
-
-const mockMessageError = jest.fn();
-jest.mock('antd', () => {
-	const actual = jest.requireActual('antd');
-	return {
-		...actual,
-		message: { error: (...args: unknown[]): void => mockMessageError(...args) },
-	};
-});
-
-const mockDownloadFile = downloadFile as jest.Mock;
-
-function timeSeriesResponse(): QueryRangeResponseV5 {
-	return {
-		type: 'time_series',
-		data: {
-			results: [
-				{
-					queryName: 'A',
-					aggregations: [
-						{
-							index: 0,
-							alias: '',
-							meta: {},
-							series: [
-								{
-									labels: [{ key: { name: 'service' }, value: 'a' }],
-									values: [{ timestamp: 1000, value: 12 }],
-								},
-							],
-						},
-					],
-				},
-			],
-		},
-		meta: {},
-	} as unknown as QueryRangeResponseV5;
-}
-
-describe('useClientExport', () => {
-	beforeEach(() => {
-		jest.clearAllMocks();
-		// Freeze the clock so filenames are deterministic — asserted against the
-		// real getTimestampedFileName (the format itself is pinned by an exact
-		// string in downloadFile.test).
-		jest.useFakeTimers().setSystemTime(new Date(2026, 6, 13, 14, 32, 5));
-	});
-
-	afterEach(() => {
-		jest.useRealTimers();
-	});
-
-	it('exports time_series as CSV to a timestamped <fileName>.csv', () => {
-		const { result } = renderHook(() =>
-			useClientExport({
-				response: timeSeriesResponse(),
-				fileName: 'chart',
-				legendMap: { A: '{{service}}' },
-			}),
-		);
-
-		act(() => {
-			result.current.handleExport({ format: ExportFormat.Csv });
-		});
-
-		expect(mockDownloadFile).toHaveBeenCalledTimes(1);
-		const [content, name, mime] = mockDownloadFile.mock.calls[0];
-		// delegation: the hook names files via getTimestampedFileName
-		expect(name).toBe(getTimestampedFileName('chart', 'csv'));
-		expect(mime).toContain('text/csv');
-		expect(content).toContain('service');
-		expect(content).toContain('a');
-	});
-
-	it('exports as JSONL to a timestamped <fileName>.jsonl with the ndjson mime', () => {
-		const { result } = renderHook(() =>
-			useClientExport({ response: timeSeriesResponse() }),
-		);
-
-		act(() => {
-			result.current.handleExport({ format: ExportFormat.Jsonl });
-		});
-
-		const [content, name, mime] = mockDownloadFile.mock.calls[0];
-		expect(name).toBe(getTimestampedFileName('export', 'jsonl'));
-		expect(mime).toContain('ndjson');
-		expect(content).toContain('"series"');
-	});
-
-	it('does nothing when there is no response', () => {
-		const { result } = renderHook(() => useClientExport({}));
-
-		act(() => {
-			result.current.handleExport({ format: ExportFormat.Csv });
-		});
-
-		expect(mockDownloadFile).not.toHaveBeenCalled();
-		expect(mockMessageError).not.toHaveBeenCalled();
-	});
-
-	it('shows an error and does not download for unsupported result types', () => {
-		const raw = {
-			type: 'raw',
-			data: { results: [] },
-			meta: {},
-		} as unknown as QueryRangeResponseV5;
-		const { result } = renderHook(() => useClientExport({ response: raw }));
-
-		act(() => {
-			result.current.handleExport({ format: ExportFormat.Csv });
-		});
-
-		expect(mockDownloadFile).not.toHaveBeenCalled();
-		expect(mockMessageError).toHaveBeenCalled();
-	});
-});

frontend/src/hooks/useExportData/useClientExport.ts

@@ -1,93 +0,0 @@
-import { message } from 'antd';
-import { REQUEST_TYPES } from 'api/v5/queryRange/constants';
-import {
-	downloadFile,
-	getTimestampedFileName,
-} from 'lib/exportData/downloadFile';
-import { exportTimeseriesData } from 'lib/exportData/exportTimeseriesData';
-import { toCsv } from 'lib/exportData/toCsv';
-import { toJsonl } from 'lib/exportData/toJsonl';
-import { ExportFormat, SerializedTable } from 'lib/exportData/types';
-import { useCallback, useState } from 'react';
-import { Query } from 'types/api/queryBuilder/queryBuilderData';
-import { QueryRangeResponseV5, TimeSeriesData } from 'types/api/v5/queryRange';
-
-const FORMAT_META: Record<ExportFormat, { mime: string; extension: string }> = {
-	[ExportFormat.Csv]: { mime: 'text/csv;charset=utf-8;', extension: 'csv' },
-	[ExportFormat.Jsonl]: {
-		mime: 'application/x-ndjson;charset=utf-8;',
-		extension: 'jsonl',
-	},
-};
-
-// Picks the serializer for the response's request type. Narrows the results
-// union via the response discriminant. scalar lands with #5591; raw/trace are
-// server-exported, distribution is never emitted.
-function serialize(
-	response: QueryRangeResponseV5,
-	yAxisUnit?: string,
-	legendMap?: Record<string, string>,
-	query?: Query,
-): SerializedTable {
-	if (response.type === REQUEST_TYPES.TIME_SERIES) {
-		return exportTimeseriesData({
-			data: response.data.results as TimeSeriesData[],
-			yAxisUnit,
-			legendMap,
-			query,
-		});
-	}
-
-	throw new Error(`Export is not supported for "${response.type}" results`);
-}
-
-interface UseClientExportProps {
-	response?: QueryRangeResponseV5;
-	query?: Query;
-	yAxisUnit?: string;
-	fileName?: string;
-	legendMap?: Record<string, string>;
-}
-
-interface ClientExportOptions {
-	format: ExportFormat;
-}
-
-interface UseClientExportReturn {
-	isExporting: boolean;
-	handleExport: (options: ClientExportOptions) => void;
-}
-
-export function useClientExport({
-	response, // currently supports only qb v5 response. Can extend to support future responses.
-	query,
-	yAxisUnit,
-	fileName = 'export',
-	legendMap,
-}: UseClientExportProps): UseClientExportReturn {
-	const [isExporting, setIsExporting] = useState<boolean>(false);
-
-	const handleExport = useCallback(
-		({ format }: ClientExportOptions): void => {
-			if (!response) {
-				return;
-			}
-
-			setIsExporting(true);
-			try {
-				const table = serialize(response, yAxisUnit, legendMap, query);
-				const content =
-					format === ExportFormat.Jsonl ? toJsonl(table) : toCsv(table);
-				const { mime, extension } = FORMAT_META[format];
-				downloadFile(content, getTimestampedFileName(fileName, extension), mime);
-			} catch {
-				message.error('Failed to export data. Please try again.');
-			} finally {
-				setIsExporting(false);
-			}
-		},
-		[response, query, yAxisUnit, fileName, legendMap],
-	);
-
-	return { isExporting, handleExport };
-}

frontend/src/lib/exportData/__tests__/downloadFile.test.ts

@@ -1,56 +0,0 @@
-import { downloadFile, getTimestampedFileName } from '../downloadFile';
-
-// jsdom doesn't implement the object-URL APIs; define stubs so jest.spyOn can wrap them.
-if (typeof URL.createObjectURL !== 'function') {
-	URL.createObjectURL = (): string => '';
-}
-if (typeof URL.revokeObjectURL !== 'function') {
-	URL.revokeObjectURL = (): void => undefined;
-}
-
-describe('downloadFile', () => {
-	afterEach(() => {
-		jest.restoreAllMocks();
-	});
-
-	it('builds a blob anchor, clicks it, and revokes the object URL', () => {
-		const click = jest.fn();
-		const remove = jest.fn();
-		const anchor = {
-			href: '',
-			download: '',
-			click,
-			remove,
-		} as unknown as HTMLAnchorElement;
-
-		(
-			jest.spyOn(document, 'createElement') as unknown as jest.Mock
-		).mockReturnValue(anchor);
-		const createObjectURL = jest
-			.spyOn(URL, 'createObjectURL')
-			.mockReturnValue('blob:mock');
-		const revokeObjectURL = jest.spyOn(URL, 'revokeObjectURL');
-
-		downloadFile('hello', 'export.csv', 'text/csv');
-
-		expect(anchor.download).toBe('export.csv');
-		expect(anchor.href).toBe('blob:mock');
-		expect(click).toHaveBeenCalledTimes(1);
-		expect(createObjectURL).toHaveBeenCalled();
-		expect(revokeObjectURL).toHaveBeenCalledWith('blob:mock');
-	});
-});
-
-describe('getTimestampedFileName', () => {
-	afterEach(() => {
-		jest.useRealTimers();
-	});
-
-	it('appends a local timestamp between base and extension', () => {
-		jest.useFakeTimers().setSystemTime(new Date(2026, 6, 8, 14, 32, 5));
-
-		expect(getTimestampedFileName('logs-timeseries', 'csv')).toBe(
-			'logs-timeseries-2026-07-08_14-32-05.csv',
-		);
-	});
-});

frontend/src/lib/exportData/__tests__/exportTimeseriesData.test.ts

@@ -1,188 +0,0 @@
-import { Query } from 'types/api/queryBuilder/queryBuilderData';
-import { TimeSeries, TimeSeriesData } from 'types/api/v5/queryRange';
-
-import { exportTimeseriesData } from '../exportTimeseriesData';
-
-const iso = (ms: number): string => new Date(ms).toISOString();
-
-function makeSeries(
-	labels: Record<string, string>,
-	values: [number, number][],
-): TimeSeries {
-	return {
-		labels: Object.entries(labels).map(([name, value]) => ({
-			key: { name },
-			value,
-		})),
-		values: values.map(([timestamp, value]) => ({ timestamp, value })),
-	};
-}
-
-function makeQuery(
-	queryName: string,
-	buckets: { index?: number; alias?: string; series: TimeSeries[] }[],
-): TimeSeriesData {
-	return {
-		queryName,
-		aggregations: buckets.map((bucket, i) => ({
-			index: bucket.index ?? i,
-			alias: bucket.alias ?? '',
-			meta: {},
-			series: bucket.series,
-		})),
-	};
-}
-
-describe('exportTimeseriesData', () => {
-	it('one row per point: query column, label columns, unit in value header, legend naming', () => {
-		const data = [
-			makeQuery('A', [
-				{
-					series: [
-						makeSeries({ service_name: 'frontend' }, [
-							[1000, 12],
-							[2000, 15],
-						]),
-					],
-				},
-			]),
-		];
-
-		const table = exportTimeseriesData({
-			data,
-			yAxisUnit: 'ms',
-			legendMap: { A: '{{service_name}}' },
-		});
-
-		expect(table.headers).toStrictEqual([
-			'timestamp',
-			'query',
-			'series',
-			'service_name',
-			'value (ms)',
-		]);
-		expect(table.rows).toStrictEqual([
-			[iso(1000), 'A', 'frontend', 'frontend', 12],
-			[iso(2000), 'A', 'frontend', 'frontend', 15],
-		]);
-	});
-
-	it('no legend falls back to the label-set name from getLabelName', () => {
-		const data = [
-			makeQuery('A', [
-				{ series: [makeSeries({ service_name: 'frontend' }, [[1000, 12]])] },
-			]),
-		];
-
-		const table = exportTimeseriesData({ data });
-
-		expect(table.rows).toStrictEqual([
-			[iso(1000), 'A', '{service_name="frontend"}', 'frontend', 12],
-		]);
-	});
-
-	it('multi-query: query is its own column; label keys are unioned', () => {
-		const data = [
-			makeQuery('A', [{ series: [makeSeries({ service: 'x' }, [[1000, 1]])] }]),
-			makeQuery('B', [{ series: [makeSeries({ service: 'y' }, [[1000, 2]])] }]),
-		];
-
-		const table = exportTimeseriesData({
-			data,
-			legendMap: { A: '{{service}}', B: '{{service}}' },
-		});
-
-		expect(table.headers).toStrictEqual([
-			'timestamp',
-			'query',
-			'series',
-			'service',
-			'value',
-		]);
-		expect(table.rows).toStrictEqual([
-			[iso(1000), 'A', 'x', 'x', 1],
-			[iso(1000), 'B', 'y', 'y', 2],
-		]);
-	});
-
-	it('multi-aggregation with the builder query: names match the chart legend', () => {
-		const data = [
-			makeQuery('A', [
-				{ index: 0, alias: '__result_0', series: [makeSeries({}, [[1000, 5]])] },
-				{
-					index: 1,
-					alias: '__result_1',
-					series: [makeSeries({}, [[1000, 300]])],
-				},
-			]),
-			makeQuery('B', [
-				{
-					index: 0,
-					alias: '__result_0',
-					series: [
-						makeSeries({ 'cloud.account.id': 'signoz-staging' }, [[1000, 7]]),
-					],
-				},
-			]),
-		];
-
-		const query = {
-			queryType: 'builder',
-			builder: {
-				queryData: [
-					{
-						queryName: 'A',
-						dataSource: 'logs',
-						aggregations: [
-							{ expression: 'count()' },
-							{ expression: 'avg(code.lineno)' },
-						],
-						groupBy: [],
-					},
-					{
-						queryName: 'B',
-						dataSource: 'logs',
-						aggregations: [{ expression: 'count()' }],
-						groupBy: [{ key: 'cloud.account.id' }],
-					},
-				],
-				queryFormulas: [],
-			},
-		} as unknown as Query;
-
-		const table = exportTimeseriesData({ data, query });
-
-		expect(table.rows).toStrictEqual([
-			[iso(1000), 'A', 'count()-A', '', 5],
-			[iso(1000), 'A', 'avg(code.lineno)-A', '', 300],
-			[iso(1000), 'B', '{cloud.account.id="signoz-staging"}', 'signoz-staging', 7],
-		]);
-	});
-
-	it('multi-aggregation without the builder query: falls back to base names', () => {
-		const data = [
-			makeQuery('A', [
-				{ index: 0, alias: '__result_0', series: [makeSeries({}, [[1000, 5]])] },
-				{
-					index: 1,
-					alias: '__result_1',
-					series: [makeSeries({}, [[1000, 300]])],
-				},
-			]),
-		];
-
-		const table = exportTimeseriesData({ data });
-
-		expect(table.rows).toStrictEqual([
-			[iso(1000), 'A', 'A', 5],
-			[iso(1000), 'A', 'A', 300],
-		]);
-	});
-
-	it('empty data: returns a headers-only table', () => {
-		expect(exportTimeseriesData({ data: [] })).toStrictEqual({
-			headers: ['timestamp', 'query', 'series', 'value'],
-			rows: [],
-		});
-	});
-});

frontend/src/lib/exportData/__tests__/formatters.test.ts

@@ -1,42 +0,0 @@
-import { toCsv } from '../toCsv';
-import { toJsonl } from '../toJsonl';
-import { SerializedTable } from '../types';
-
-const table: SerializedTable = {
-	headers: ['timestamp', 'value'],
-	rows: [
-		['t1', 12],
-		['t2', 15],
-	],
-};
-
-describe('toCsv', () => {
-	it('emits a header row then one row per record, in column order', () => {
-		expect(toCsv(table).split(/\r?\n/)).toStrictEqual([
-			'timestamp,value',
-			't1,12',
-			't2,15',
-		]);
-	});
-
-	it('quotes values containing the delimiter', () => {
-		const csv = toCsv({ headers: ['name', 'value'], rows: [['a,b', 1]] });
-		expect(csv.split(/\r?\n/)).toStrictEqual(['name,value', '"a,b",1']);
-	});
-
-	it('emits only the header row when there are no data rows', () => {
-		expect(toCsv({ headers: ['timestamp'], rows: [] })).toBe('timestamp\r\n');
-	});
-});
-
-describe('toJsonl', () => {
-	it('emits one JSON object per row keyed by header', () => {
-		expect(toJsonl(table)).toBe(
-			'{"timestamp":"t1","value":12}\n{"timestamp":"t2","value":15}',
-		);
-	});
-
-	it('emits an empty string when there are no rows', () => {
-		expect(toJsonl({ headers: ['timestamp'], rows: [] })).toBe('');
-	});
-});

frontend/src/lib/exportData/downloadFile.ts

@@ -1,29 +0,0 @@
-/** Triggers a browser download of in-memory string content as a file. */
-export function downloadFile(
-	content: string,
-	fileName: string,
-	mime: string,
-): void {
-	const blob = new Blob([content], { type: mime });
-	const url = URL.createObjectURL(blob);
-	const link = document.createElement('a');
-	link.href = url;
-	link.download = fileName;
-	link.click();
-	link.remove();
-	URL.revokeObjectURL(url);
-}
-
-/** `base` + local timestamp + extension, e.g. `logs-timeseries-2026-07-08_14-32-05.csv`.
- * Keeps repeated exports from colliding and records when the export was taken. */
-export function getTimestampedFileName(
-	base: string,
-	extension: string,
-): string {
-	const now = new Date();
-	const pad = (value: number): string => String(value).padStart(2, '0');
-	const stamp = `${now.getFullYear()}-${pad(now.getMonth() + 1)}-${pad(
-		now.getDate(),
-	)}_${pad(now.getHours())}-${pad(now.getMinutes())}-${pad(now.getSeconds())}`;
-	return `${base}-${stamp}.${extension}`;
-}

frontend/src/lib/exportData/exportTimeseriesData.ts

@@ -1,154 +0,0 @@
-import { getLegend } from 'lib/dashboard/getQueryResults';
-import getLabelName from 'lib/getLabelName';
-import { Query } from 'types/api/queryBuilder/queryBuilderData';
-import { TimeSeries, TimeSeriesData } from 'types/api/v5/queryRange';
-import { QueryData } from 'types/api/widgets/getQuery';
-
-import { SerializedTable } from './types';
-
-interface ExportTimeseriesDataArgs {
-	data: TimeSeriesData[];
-	yAxisUnit?: string;
-	legendMap?: Record<string, string>;
-	// The builder query that produced the data — lets series names resolve
-	// aggregation aliases/expressions exactly like the chart legend does.
-	query?: Query;
-}
-
-// One row of the flattened V5 tree: a single (query, aggregation, label-set) series.
-interface FlatSeries {
-	queryName: string;
-	labels: Record<string, string>;
-	name: string;
-	values: { timestamp: number; value: number }[];
-}
-
-// V5 labels [{key:{name}, value}] → {name: value} (the getLabelName contract).
-function foldLabels(labels: TimeSeries['labels']): Record<string, string> {
-	const record: Record<string, string> = {};
-	(labels ?? []).forEach((label) => {
-		if (label.key?.name) {
-			record[label.key.name] = String(label.value);
-		}
-	});
-	return record;
-}
-
-// Series display name, matching the chart legend: getLabelName for the base
-// (legend template / label-set), then getLegend to resolve the aggregation
-// alias/expression from the builder query (the response only carries
-// auto-generated `__result_N` aliases). Same chain the uPlot layer uses.
-function seriesName(args: {
-	labels: Record<string, string>;
-	queryName: string;
-	legend: string;
-	aggIndex: number;
-	alias: string;
-	query?: Query;
-}): string {
-	const { labels, queryName, legend, aggIndex, alias, query } = args;
-	const baseName = getLabelName(labels, queryName, legend);
-
-	if (!query) {
-		return baseName;
-	}
-
-	const legacySeries = {
-		queryName,
-		metric: labels,
-		values: [],
-		metaData: { alias, index: aggIndex, queryName },
-	} as QueryData;
-
-	return getLegend(legacySeries, query, baseName);
-}
-
-// Walk results → aggregations → series into a flat, named list.
-function flatten(
-	data: TimeSeriesData[],
-	legendMap?: Record<string, string>,
-	query?: Query,
-): FlatSeries[] {
-	const flat: FlatSeries[] = [];
-	data.forEach((result) => {
-		const queryName = result.queryName ?? '';
-		const legend = legendMap?.[queryName] ?? '';
-		(result.aggregations ?? []).forEach((bucket) => {
-			(bucket.series ?? []).forEach((series) => {
-				const labels = foldLabels(series.labels);
-				flat.push({
-					queryName,
-					labels,
-					name: seriesName({
-						labels,
-						queryName,
-						legend,
-						aggIndex: bucket.index ?? 0,
-						alias: bucket.alias ?? '',
-						query,
-					}),
-					values: (series.values ?? []).map((value) => ({
-						timestamp: value.timestamp,
-						value: value.value,
-					})),
-				});
-			});
-		});
-	});
-	return flat;
-}
-
-// Appends the y-axis unit to the value header: `value` → `value (ms)`.
-function withUnit(header: string, yAxisUnit?: string): string {
-	return yAxisUnit ? `${header} (${yAxisUnit})` : header;
-}
-
-function toIso(timestamp: number): string {
-	return new Date(timestamp).toISOString();
-}
-
-// Tidy (LONG) layout: one row per (series, timestamp). query is its own column.
-function buildTable(flat: FlatSeries[], yAxisUnit?: string): SerializedTable {
-	const labelKeySet = new Set<string>();
-	flat.forEach((series) => {
-		Object.keys(series.labels).forEach((key) => labelKeySet.add(key));
-	});
-	const labelKeys = Array.from(labelKeySet).sort();
-
-	const headers = [
-		'timestamp',
-		'query',
-		'series',
-		...labelKeys,
-		withUnit('value', yAxisUnit),
-	];
-
-	const rows: (string | number)[][] = [];
-	flat.forEach((series) => {
-		series.values.forEach(({ timestamp, value }) => {
-			rows.push([
-				toIso(timestamp),
-				series.queryName,
-				series.name,
-				...labelKeys.map((key) => series.labels[key] ?? ''),
-				value,
-			]);
-		});
-	});
-
-	return { headers, rows };
-}
-
-/**
- * Serializes a V5 time_series result into a format-agnostic tidy table — one
- * row per (series, timestamp), labels as columns, raw values.
- * Pure — walks the V5 tree directly; series names match the chart legend.
- */
-export function exportTimeseriesData({
-	data,
-	yAxisUnit,
-	legendMap,
-	query,
-}: ExportTimeseriesDataArgs): SerializedTable {
-	return buildTable(flatten(data, legendMap, query), yAxisUnit);
-}

frontend/src/lib/exportData/toCsv.ts

@@ -1,8 +0,0 @@
-import { unparse } from 'papaparse';
-
-import { SerializedTable } from './types';
-
-/** Serializes a table to CSV. `fields` pins column order regardless of row keys. */
-export function toCsv(table: SerializedTable): string {
-	return unparse({ fields: table.headers, data: table.rows });
-}

frontend/src/lib/exportData/toJsonl.ts

@@ -1,12 +0,0 @@
-import { SerializedTable } from './types';
-
-/** Serializes a table to newline-delimited JSON: one object per row, keyed by header. */
-export function toJsonl(table: SerializedTable): string {
-	return table.rows
-		.map((row) =>
-			JSON.stringify(
-				Object.fromEntries(table.headers.map((header, i) => [header, row[i]])),
-			),
-		)
-		.join('\n');
-}

frontend/src/lib/exportData/types.ts

@@ -1,13 +0,0 @@
-/** Format-agnostic tabular result produced by every exporter. Consumed by the
- * CSV/JSONL formatters */
-export interface SerializedTable {
-	headers: string[];
-	// One entry per header, in header order. Empty string marks a gap.
-	rows: (string | number)[][];
-}
-
-/** File formats a client-side export can be downloaded as. */
-export enum ExportFormat {
-	Csv = 'csv',
-	Jsonl = 'jsonl',
-}

frontend/src/pages/DashboardPageV2/DashboardContainer/PanelEditor/hooks/__tests__/useSwitchToViewMode.test.ts

@@ -1,10 +1,5 @@
 import { renderHook } from '@testing-library/react';
-import type { DashboardtypesPanelSpecDTO } from 'api/generated/services/sigNoz.schemas';
 import { PANEL_TYPES } from 'constants/queryBuilder';
-import {
-	clearViewPanelHandoff,
-	readViewPanelHandoff,
-} from 'pages/DashboardPageV2/DashboardContainer/PanelsAndSectionsLayout/Panel/ViewPanelModal/viewPanelHandoffStore';
 import type { Query } from 'types/api/queryBuilder/queryBuilderData';
 
 import { useSwitchToViewMode } from '../useSwitchToViewMode';
@@ -23,16 +18,11 @@ jest.mock('hooks/useUrlQuery', () => ({
 }));
 
 const query = { queryType: 'builder' } as unknown as Query;
-const spec = {
-	plugin: { kind: 'signoz/TimeSeriesPanel' },
-	display: { name: 'CPU' },
-} as unknown as DashboardtypesPanelSpecDTO;
 
 describe('useSwitchToViewMode', () => {
 	beforeEach(() => {
 		jest.clearAllMocks();
 		mockSearch = '';
-		clearViewPanelHandoff();
 	});
 
 	function invoke(): void {
@@ -42,7 +32,6 @@ describe('useSwitchToViewMode', () => {
 				panelId: 'panel-1',
 				panelType: PANEL_TYPES.TIME_SERIES,
 				query,
-				spec,
 			}),
 		);
 		result.current();
@@ -63,21 +52,6 @@ describe('useSwitchToViewMode', () => {
 		).toStrictEqual(query);
 	});
 
-	it('stashes the live draft spec in the sessionStorage handoff, not the URL', () => {
-		invoke();
-
-		expect(readViewPanelHandoff('dash-1', 'panel-1')).toStrictEqual(spec);
-		// The spec must not bloat the URL — the config-only display name never leaks into it.
-		expect(mockSafeNavigate.mock.calls[0][0]).not.toContain('CPU');
-	});
-
-	it('scopes the handoff to the exact dashboard + panel', () => {
-		invoke();
-
-		expect(readViewPanelHandoff('dash-1', 'other-panel')).toBeNull();
-		expect(readViewPanelHandoff('other-dash', 'panel-1')).toBeNull();
-	});
-
 	it('carries dashboard variables through and drops other editor URL state', () => {
 		mockSearch = 'variables=%7B%22a%22%3A1%7D&compositeQuery=stale';
 		invoke();

frontend/src/pages/DashboardPageV2/DashboardContainer/PanelEditor/hooks/useSwitchToViewMode.ts

@@ -1,6 +1,5 @@
 import { useCallback } from 'react';
 import { generatePath } from 'react-router-dom';
-import type { DashboardtypesPanelSpecDTO } from 'api/generated/services/sigNoz.schemas';
 import { QueryParams } from 'constants/query';
 import type { PANEL_TYPES } from 'constants/queryBuilder';
 import ROUTES from 'constants/routes';
@@ -8,35 +7,27 @@ import { useSafeNavigate } from 'hooks/useSafeNavigate';
 import useUrlQuery from 'hooks/useUrlQuery';
 import type { Query } from 'types/api/queryBuilder/queryBuilderData';
 
-import { writeViewPanelHandoff } from '../../PanelsAndSectionsLayout/Panel/ViewPanelModal/viewPanelHandoffStore';
-
 interface UseSwitchToViewModeArgs {
 	dashboardId: string;
 	panelId: string;
 	panelType: PANEL_TYPES;
 	query: Query;
-	/** Live (un-saved) draft spec — the query rides in the URL, the rest via the handoff. */
-	spec: DashboardtypesPanelSpecDTO;
 }
 
 /**
- * Leaves the editor for the dashboard with this panel expanded in the View modal, seeded with
- * the live (un-saved) query + config — V1's "Switch to View Mode". The query rides in the URL
- * (`compositeQuery`); the rest of the spec rides in a tab-scoped sessionStorage handoff.
+ * Callback that leaves the editor for the dashboard with this panel expanded in the
+ * View modal, seeded with the live (un-saved) query — V1's "Switch to View Mode".
  */
 export function useSwitchToViewMode({
 	dashboardId,
 	panelId,
 	panelType,
 	query,
-	spec,
 }: UseSwitchToViewModeArgs): () => void {
 	const { safeNavigate } = useSafeNavigate();
 	const urlQuery = useUrlQuery();
 
 	return useCallback((): void => {
-		writeViewPanelHandoff({ dashboardId, panelId, spec });
-
 		const params = new URLSearchParams();
 		const variables = urlQuery.get(QueryParams.variables);
 		if (variables) {
@@ -51,5 +42,5 @@ export function useSwitchToViewMode({
 		safeNavigate(
 			`${generatePath(ROUTES.DASHBOARD, { dashboardId })}?${params.toString()}`,
 		);
-	}, [safeNavigate, urlQuery, dashboardId, panelId, panelType, query, spec]);
+	}, [safeNavigate, urlQuery, dashboardId, panelId, panelType, query]);
 }

frontend/src/pages/DashboardPageV2/DashboardContainer/PanelEditor/index.tsx

@@ -38,8 +38,6 @@ import { useTableColumns } from './hooks/useTableColumns';
 import ListColumnsEditor from './ListColumnsEditor/ListColumnsEditor';
 
 import styles from './PanelEditor.module.scss';
-import logEvent from '@/api/common/logEvent';
-import { DashboardEvents } from '../../constants/events';
 
 interface PanelEditorContainerProps {
 	dashboardId: string;
@@ -206,7 +204,6 @@ function PanelEditorContainer({
 		panelId,
 		panelType: PANEL_KIND_TO_PANEL_TYPE[panelKind],
 		query: currentQuery,
-		spec: draft.spec,
 	});
 
 	const setScrollTargetId = useScrollIntoViewStore((s) => s.setScrollTargetId);
@@ -237,13 +234,6 @@ function PanelEditorContainer({
 		onClose();
 	}, [isNew, panelId, setScrollTargetId, onClose]);
 
-	const switchToViewMode = useCallback((): void => {
-		logEvent(DashboardEvents.SWITCH_TO_VIEW_MODE, {
-			panelId: panelId,
-		});
-		onSwitchToView();
-	}, [onSwitchToView]);
-
 	return (
 		<div className={styles.page} data-testid="panel-editor-v2">
 			<Header
@@ -253,7 +243,7 @@ function PanelEditorContainer({
 				readOnly={!isEditable}
 				readOnlyReason={editDisabledReason}
 				onSave={onSave}
-				onSwitchToView={switchToViewMode}
+				onSwitchToView={onSwitchToView}
 				onClose={onCloseEditor}
 			/>
 			<ResizablePanelGroup

frontend/src/pages/DashboardPageV2/DashboardContainer/PanelsAndSectionsLayout/Panel/ViewPanelModal/ViewPanelModalContent.tsx

@@ -16,8 +16,6 @@ import ViewPanelModalHeader from './ViewPanelModalHeader';
 import { useViewPanelMode } from './useViewPanelMode';
 import { useViewPanelTimeWindow } from './useViewPanelTimeWindow';
 import styles from './ViewPanelModal.module.scss';
-import logEvent from 'api/common/logEvent';
-import { DashboardEvents } from 'pages/DashboardPageV2/constants/events';
 
 interface ViewPanelModalContentProps {
 	panel: DashboardtypesPanelDTO;
@@ -99,14 +97,6 @@ function ViewPanelModalContent({
 		return null;
 	}
 
-	const onSwitchToEdit = (): void => {
-		// Carry the drilldown edits so the editor opens on them, not the saved panel.
-		logEvent(DashboardEvents.SWITCH_TO_EDIT_MODE, {
-			panelId: panelId,
-		});
-		openPanelEditor(panelId, { editSpec: buildSaveSpec(draft.spec) });
-	};
-
 	return (
 		<div className={styles.content} data-testid="view-panel-modal-content">
 			<ViewPanelModalHeader
@@ -124,7 +114,10 @@ function ViewPanelModalContent({
 						refreshWindow();
 					}
 				}}
-				onSwitchToEdit={onSwitchToEdit}
+				onSwitchToEdit={(): void =>
+					// Carry the drilldown edits so the editor opens on them, not the saved panel.
+					openPanelEditor(panelId, { editSpec: buildSaveSpec(draft.spec) })
+				}
 				panelKind={draft.spec.plugin.kind}
 				queryType={queryType}
 				signal={signal}

frontend/src/pages/DashboardPageV2/DashboardContainer/PanelsAndSectionsLayout/Panel/ViewPanelModal/ViewPanelModalHeader.tsx

@@ -13,7 +13,6 @@ import type { PanelKind } from 'pages/DashboardPageV2/DashboardContainer/Panels/
 import type { EQueryType } from 'types/common/dashboard';
 
 import styles from './ViewPanelModal.module.scss';
-import { useDashboardStore } from 'pages/DashboardPageV2/DashboardContainer/store/useDashboardStore';
 
 interface ViewPanelModalHeaderProps {
 	selectedInterval: Time | CustomTimeType;
@@ -65,10 +64,6 @@ function ViewPanelModalHeader({
 	// Same capabilities-guarded options as the editor's PanelTypeSwitcher, so the two
 	// selectors disable the same kinds (e.g. List under PromQL, metrics-only kinds).
 	const panelTypeItems = usePanelTypeSelectItems({ queryType, signal });
-	const canEditDashboard = useDashboardStore((s) => s.canEditDashboard);
-	const isLocked = useDashboardStore((s) => s.isLocked);
-
-	const canSwitchToEdit = canEditDashboard && !isLocked;
 
 	return (
 		<div className={styles.toolbar}>
@@ -80,17 +75,15 @@ function ViewPanelModalHeader({
 					onChange={onChangePanelKind}
 				/>
 			</div>
-			{canSwitchToEdit && (
-				<Button
-					variant="outlined"
-					color="secondary"
-					prefix={<PenLine />}
-					onClick={onSwitchToEdit}
-					data-testid="view-panel-switch-to-edit"
-				>
-					Switch to Edit Mode
-				</Button>
-			)}
+			<Button
+				variant="outlined"
+				color="secondary"
+				prefix={<PenLine />}
+				onClick={onSwitchToEdit}
+				data-testid="view-panel-switch-to-edit"
+			>
+				Switch to Edit Mode
+			</Button>
 			<Button
 				variant="link"
 				color="primary"

frontend/src/pages/DashboardPageV2/DashboardContainer/PanelsAndSectionsLayout/Panel/ViewPanelModal/useViewPanelMode.ts

@@ -23,11 +23,8 @@ import {
 	type PanelQueryTimeOverride,
 	type UsePanelQueryResult,
 } from 'pages/DashboardPageV2/DashboardContainer/hooks/usePanelQuery';
-import { useDashboardStore } from 'pages/DashboardPageV2/DashboardContainer/store/useDashboardStore';
 import type { EQueryType } from 'types/common/dashboard';
 
-import { readViewPanelHandoff } from './viewPanelHandoffStore';
-
 interface UseViewPanelModeArgs {
 	panel: DashboardtypesPanelDTO;
 	panelId: string;
@@ -80,33 +77,25 @@ export function useViewPanelMode({
 }: UseViewPanelModeArgs): UseViewPanelModeReturn {
 	const { currentQuery, redirectWithQueryBuilderData } = useQueryBuilder();
 
-	// Config edits from the editor's "Switch to View Mode" arrive via the handoff; the query
-	// still comes from the URL. Falls back to the saved panel for a plain grid "View".
-	const dashboardId = useDashboardStore((s) => s.dashboardId);
-	const baseSpec = useMemo<DashboardtypesPanelSpecDTO>(
-		() => readViewPanelHandoff(dashboardId, panelId) ?? panel.spec,
-		// eslint-disable-next-line react-hooks/exhaustive-deps -- mount-only seed
-		[],
-	);
-
-	// Mount-only so a refresh re-seeds and in-modal edits survive (V1 parity).
-	const compositeQuery = useGetCompositeQueryParam();
+	// Seed the draft from the URL (`compositeQuery` + `graphType`) when present, else the saved
+	// panel — mount-only, so a refresh re-seeds from the URL and in-modal edits survive (V1 parity).
+	const urlQuery = useGetCompositeQueryParam();
 	const urlGraphType = useUrlQuery().get(
 		QueryParams.graphType,
 	) as PANEL_TYPES | null;
 	const initialPanel = useMemo<DashboardtypesPanelDTO>(
 		() =>
-			compositeQuery
+			urlQuery
 				? {
 						...panel,
 						spec: buildViewPanelSpec({
-							spec: baseSpec,
-							query: compositeQuery,
+							spec: panel.spec,
+							query: urlQuery,
 							panelType:
-								urlGraphType ?? PANEL_KIND_TO_PANEL_TYPE[baseSpec.plugin.kind],
+								urlGraphType ?? PANEL_KIND_TO_PANEL_TYPE[panel.spec.plugin.kind],
 						}),
 					}
-				: { ...panel, spec: baseSpec },
+				: panel,
 		// eslint-disable-next-line react-hooks/exhaustive-deps -- mount-only seed from the URL
 		[],
 	);

frontend/src/pages/DashboardPageV2/DashboardContainer/PanelsAndSectionsLayout/Panel/ViewPanelModal/viewPanelHandoffStore.ts

@@ -1,43 +0,0 @@
-import getSessionStorage from 'api/browser/sessionstorage/get';
-import removeSessionStorage from 'api/browser/sessionstorage/remove';
-import setSessionStorage from 'api/browser/sessionstorage/set';
-import type { DashboardtypesPanelSpecDTO } from 'api/generated/services/sigNoz.schemas';
-import { SESSIONSTORAGE } from 'constants/sessionStorage';
-
-interface ViewPanelHandoff {
-	/** Correlator: the read returns the spec only for this exact dashboard + panel. */
-	dashboardId: string;
-	panelId: string;
-	spec: DashboardtypesPanelSpecDTO;
-}
-
-/**
- * Tab-scoped handoff of the editor's un-saved draft spec to the View modal, so "Switch to View
- * Mode" carries config edits — not just the query, which stays in the URL. sessionStorage keeps
- * the link small yet survives a refresh, and clears the edits when the tab closes.
- */
-export function writeViewPanelHandoff(handoff: ViewPanelHandoff): void {
-	setSessionStorage(SESSIONSTORAGE.VIEW_PANEL_HANDOFF, JSON.stringify(handoff));
-}
-
-export function readViewPanelHandoff(
-	dashboardId: string,
-	panelId: string,
-): DashboardtypesPanelSpecDTO | null {
-	const raw = getSessionStorage(SESSIONSTORAGE.VIEW_PANEL_HANDOFF);
-	if (!raw) {
-		return null;
-	}
-	try {
-		const handoff = JSON.parse(raw) as ViewPanelHandoff;
-		return handoff.dashboardId === dashboardId && handoff.panelId === panelId
-			? handoff.spec
-			: null;
-	} catch {
-		return null;
-	}
-}
-
-export function clearViewPanelHandoff(): void {
-	removeSessionStorage(SESSIONSTORAGE.VIEW_PANEL_HANDOFF);
-}

frontend/src/pages/DashboardPageV2/DashboardContainer/PanelsAndSectionsLayout/Panel/hooks/useViewPanel.ts

@@ -6,8 +6,6 @@ import { useSafeNavigate } from 'hooks/useSafeNavigate';
 import useUrlQuery from 'hooks/useUrlQuery';
 import type { Query } from 'types/api/queryBuilder/queryBuilderData';
 
-import { clearViewPanelHandoff } from '../ViewPanelModal/viewPanelHandoffStore';
-
 export interface UseViewPanelApi {
 	/** Panel id currently expanded in the View modal; null when none is open. */
 	expandedPanelId: string | null;
@@ -43,11 +41,10 @@ export function useViewPanel(): UseViewPanelApi {
 			// Copy before mutating: useUrlQuery returns a memoized instance.
 			const next = new URLSearchParams(urlQuery);
 			next.set(QueryParams.expandedWidgetId, panelId);
-			// Drop leftover in-modal query/kind + the editor's handoff so a plain View opens
-			// on the saved panel, not stale state the modal would otherwise hydrate from.
+			// Drop any leftover in-modal query/kind so a plain View opens on the saved
+			// panel, not a stale URL query the modal would otherwise hydrate from.
 			next.delete(QueryParams.compositeQuery);
 			next.delete(QueryParams.graphType);
-			clearViewPanelHandoff();
 			safeNavigate(`${pathname}?${next.toString()}`);
 		},
 		[pathname, safeNavigate, urlQuery],
@@ -58,8 +55,6 @@ export function useViewPanel(): UseViewPanelApi {
 			const next = new URLSearchParams(urlQuery);
 			next.set(QueryParams.expandedWidgetId, panelId);
 			next.set(QueryParams.graphType, panelType);
-			// A grid drilldown opens on the saved panel, never a stale editor handoff.
-			clearViewPanelHandoff();
 			// Same encoding the query builder uses (see `useGetCompositeQueryParam`): the URL
 			// value is `encodeURIComponent(JSON.stringify(query))`, decoded once on read.
 			next.set(
@@ -78,7 +73,6 @@ export function useViewPanel(): UseViewPanelApi {
 		// (the in-modal query builder writes compositeQuery, V1 parity).
 		next.delete(QueryParams.compositeQuery);
 		next.delete(QueryParams.graphType);
-		clearViewPanelHandoff();
 		const search = next.toString();
 		safeNavigate(search ? `${pathname}?${search}` : pathname);
 	}, [pathname, safeNavigate, urlQuery]);

frontend/src/pages/DashboardPageV2/DashboardContainer/VariablesBar/selectors/DynamicSelector.tsx

@@ -3,10 +3,7 @@ import { useQuery } from 'react-query';
 // eslint-disable-next-line no-restricted-imports
 import { useSelector } from 'react-redux';
 import { getFieldValues } from 'api/dynamicVariables/getFieldValues';
-import {
-	DASHBOARD_CACHE_TIME,
-	DASHBOARD_CACHE_TIME_ON_REFRESH_ENABLED,
-} from 'constants/queryCacheTime';
+import { DASHBOARD_CACHE_TIME } from 'constants/queryCacheTime';
 import type { AppState } from 'store/reducers';
 import type { GlobalReducer } from 'types/reducer/globalTime';
 
@@ -51,10 +48,9 @@ function DynamicSelector({
 	onChange,
 	onAutoSelect,
 }: DynamicSelectorProps): JSX.Element {
-	const { minTime, maxTime, isAutoRefreshDisabled } = useSelector<
-		AppState,
-		GlobalReducer
-	>((state) => state.globalTime);
+	const { minTime, maxTime } = useSelector<AppState, GlobalReducer>(
+		(state) => state.globalTime,
+	);
 
 	const existingQuery = useMemo(
 		() => buildExistingDynamicVariableQuery(variables, selections, variable.name),
@@ -100,10 +96,8 @@ function DynamicSelector({
 				!!variable.dynamicAttribute &&
 				(isVariableFetching || (isVariableSettled && hasVariableFetchedOnce)),
 			refetchOnWindowFocus: false,
-			// Each cycle mints a fresh key; 0 under auto-refresh so entries don't pile up (V1 parity).
-			cacheTime: isAutoRefreshDisabled
-				? DASHBOARD_CACHE_TIME
-				: DASHBOARD_CACHE_TIME_ON_REFRESH_ENABLED,
+			// Each cycle mints a fresh key; a small cacheTime bounds cache churn.
+			cacheTime: DASHBOARD_CACHE_TIME,
 			onSettled: (_, error) =>
 				error
 					? onVariableFetchFailure(variable.name)

frontend/src/pages/DashboardPageV2/DashboardContainer/VariablesBar/selectors/QuerySelector.tsx

@@ -3,10 +3,7 @@ import { useQuery } from 'react-query';
 // eslint-disable-next-line no-restricted-imports
 import { useSelector } from 'react-redux';
 import dashboardVariablesQuery from 'api/dashboard/variables/dashboardVariablesQuery';
-import {
-	DASHBOARD_CACHE_TIME,
-	DASHBOARD_CACHE_TIME_ON_REFRESH_ENABLED,
-} from 'constants/queryCacheTime';
+import { DASHBOARD_CACHE_TIME } from 'constants/queryCacheTime';
 import type { AppState } from 'store/reducers';
 import type { GlobalReducer } from 'types/reducer/globalTime';
 
@@ -47,10 +44,9 @@ function QuerySelector({
 	onChange,
 	onAutoSelect,
 }: QuerySelectorProps): JSX.Element {
-	const { minTime, maxTime, isAutoRefreshDisabled } = useSelector<
-		AppState,
-		GlobalReducer
-	>((state) => state.globalTime);
+	const { minTime, maxTime } = useSelector<AppState, GlobalReducer>(
+		(state) => state.globalTime,
+	);
 	const payload = useMemo(() => selectionToPayload(selections), [selections]);
 
 	const {
@@ -84,10 +80,8 @@ function QuerySelector({
 		{
 			enabled: isVariableFetching || (isVariableSettled && hasVariableFetchedOnce),
 			refetchOnWindowFocus: false,
-			// Each cycle mints a fresh key; 0 under auto-refresh so entries don't pile up (V1 parity).
-			cacheTime: isAutoRefreshDisabled
-				? DASHBOARD_CACHE_TIME
-				: DASHBOARD_CACHE_TIME_ON_REFRESH_ENABLED,
+			// Each cycle mints a fresh key; a small cacheTime bounds cache churn.
+			cacheTime: DASHBOARD_CACHE_TIME,
 			onSettled: (_, error) =>
 				error
 					? onVariableFetchFailure(variable.name)

frontend/src/pages/DashboardPageV2/DashboardContainer/hooks/__tests__/usePanelQuery.test.tsx

@@ -2,10 +2,6 @@
 import { useSelector } from 'react-redux';
 import { act, renderHook } from '@testing-library/react';
 import type { DashboardtypesPanelDTO } from 'api/generated/services/sigNoz.schemas';
-import {
-	DASHBOARD_CACHE_TIME,
-	DASHBOARD_CACHE_TIME_ON_REFRESH_ENABLED,
-} from 'constants/queryCacheTime';
 
 import { usePanelQuery } from '../usePanelQuery';
 import { useGetQueryRangeV5 } from '../useGetQueryRangeV5';
@@ -436,28 +432,4 @@ describe('usePanelQuery', () => {
 			expect(result.current.pagination?.pageIndex).toBe(0);
 		});
 	});
-
-	describe('cacheTime (auto-refresh OOM guard)', () => {
-		const withAutoRefreshDisabled = (disabled: boolean): void => {
-			mockUseSelector.mockImplementation((selector: unknown) =>
-				(selector as (state: { globalTime: unknown }) => unknown)({
-					globalTime: { ...DEFAULT_GLOBAL_TIME, isAutoRefreshDisabled: disabled },
-				}),
-			);
-		};
-
-		it('caches for DASHBOARD_CACHE_TIME when auto-refresh is disabled', () => {
-			withAutoRefreshDisabled(true);
-			renderHook(() => usePanelQuery({ panel: builderPanel(), panelId: 'p1' }));
-			const [{ cacheTime }] = mockUseGetQueryRangeV5.mock.calls[0];
-			expect(cacheTime).toBe(DASHBOARD_CACHE_TIME);
-		});
-
-		it('drops cacheTime to 0 when auto-refresh is enabled', () => {
-			withAutoRefreshDisabled(false);
-			renderHook(() => usePanelQuery({ panel: builderPanel(), panelId: 'p1' }));
-			const [{ cacheTime }] = mockUseGetQueryRangeV5.mock.calls[0];
-			expect(cacheTime).toBe(DASHBOARD_CACHE_TIME_ON_REFRESH_ENABLED);
-		});
-	});
 });

frontend/src/pages/DashboardPageV2/DashboardContainer/hooks/useGetQueryRangeV5.ts

@@ -13,8 +13,6 @@ export interface UseGetQueryRangeV5Args {
 	enabled: boolean;
 	/** Retain prior data across a key change (list paging) so the table + pager stay mounted. */
 	keepPreviousData?: boolean;
-	/** Unused-entry TTL; callers drop to 0 under auto-refresh to bound cache growth (V1 parity). */
-	cacheTime?: number;
 }
 
 /**
@@ -48,7 +46,6 @@ export function useGetQueryRangeV5({
 	queryKey,
 	enabled,
 	keepPreviousData,
-	cacheTime,
 }: UseGetQueryRangeV5Args): UseQueryResult<QueryRangeV5200, Error> {
 	return useQuery<QueryRangeV5200, Error>({
 		queryKey,
@@ -56,6 +53,5 @@ export function useGetQueryRangeV5({
 		enabled,
 		retry: retryUnlessClientError,
 		keepPreviousData,
-		cacheTime,
 	});
 }

frontend/src/pages/DashboardPageV2/DashboardContainer/hooks/usePanelQuery.ts

@@ -4,10 +4,6 @@ import { useQueryClient } from 'react-query';
 import { useSelector } from 'react-redux';
 import type { DashboardtypesPanelDTO } from 'api/generated/services/sigNoz.schemas';
 import { PANEL_TYPES } from 'constants/queryBuilder';
-import {
-	DASHBOARD_CACHE_TIME,
-	DASHBOARD_CACHE_TIME_ON_REFRESH_ENABLED,
-} from 'constants/queryCacheTime';
 import { REACT_QUERY_KEY } from 'constants/reactQueryKeys';
 import { AppState } from 'store/reducers';
 import { GlobalReducer } from 'types/reducer/globalTime';
@@ -114,7 +110,6 @@ export function usePanelQuery({
 		selectedTime: globalSelectedInterval,
 		maxTime,
 		minTime,
-		isAutoRefreshDisabled,
 	} = useSelector<AppState, GlobalReducer>((state) => state.globalTime);
 
 	// Resolved variable values for this dashboard, published by useResolvedVariables.
@@ -248,10 +243,6 @@ export function usePanelQuery({
 		enabled: enabled && runnable && !isWaitingOnVariable,
 		// Hold the current page while the next loads (offset re-keys) so the pager doesn't flash.
 		keepPreviousData: isPaginated,
-		// 0 under auto-refresh so time-keyed entries don't accumulate and OOM the tab (V1 parity).
-		cacheTime: isAutoRefreshDisabled
-			? DASHBOARD_CACHE_TIME
-			: DASHBOARD_CACHE_TIME_ON_REFRESH_ENABLED,
 	});
 
 	const queryClient = useQueryClient();

frontend/src/pages/DashboardPageV2/constants/events.ts

@@ -1,4 +0,0 @@
-export enum DashboardEvents {
-	SWITCH_TO_EDIT_MODE = 'View Panel: Switch to edit mode',
-	SWITCH_TO_VIEW_MODE = 'Edit Panel: Switch to view mode',
-}

frontend/src/pages/TraceDetailsV3/TraceDetailsHeader/MissingSpansBanner.tsx

@@ -5,7 +5,7 @@ import { ArrowUpRight } from '@signozhq/icons';
 import styles from './MissingSpansBanner.module.scss';
 
 const MISSING_SPANS_DOCS_URL =
-	'https://signoz.io/docs/traces-management/troubleshooting/faqs/#q-why-are-some-spans-missing-from-a-trace';
+	'https://signoz.io/docs/userguide/traces/#missing-spans';
 
 function MissingSpansBanner(): JSX.Element | null {
 	// Session-only dismissal — not persisted, so the banner returns on reload.

frontend/src/utils/docLinks.ts

@@ -14,8 +14,7 @@ const DOCLINKS = {
 		'https://signoz.io/docs/userguide/logs_clickhouse_queries/',
 	QUERY_CLICKHOUSE_METRICS:
 		'https://signoz.io/docs/userguide/write-a-metrics-clickhouse-query/',
-	AGENT_SKILL_INSTALL:
-		'https://signoz.io/docs/ai/agent-skills/#install-the-plugin',
+	AGENT_SKILL_INSTALL: 'https://signoz.io/docs/ai/agent-skills/#installation',
 };
 
 export default DOCLINKS;

pkg/apiserver/signozapiserver/spanmapper.go

@@ -98,7 +98,7 @@ func (provider *provider) addSpanMapperRoutes(router *mux.Router) error {
 			Description:         "Returns all mappers belonging to a mapping group.",
 			Request:             nil,
 			RequestContentType:  "",
-			Response:            new(spantypes.GettableSpanMappers),
+			Response:            new(spantypes.GettableSpanMapperGroups),
 			ResponseContentType: "application/json",
 			SuccessStatusCode:   http.StatusOK,
 			ErrorStatusCodes:    []int{http.StatusBadRequest, http.StatusNotFound},

pkg/flagger/registry.go

@@ -15,6 +15,8 @@ var (
 	FeatureEnableAIObservability  = featuretypes.MustNewName("enable_ai_observability")
 	FeatureEnableMetricsReduction = featuretypes.MustNewName("enable_metrics_reduction")
 	FeatureUseInfraMonitoringV2   = featuretypes.MustNewName("use_infra_monitoring_v2")
+
+	FeatureUsePrometheusClickhouseV2 = featuretypes.MustNewName("use_prometheus_clickhouse_v2")
 )
 
 func MustNewRegistry() featuretypes.Registry {
@@ -115,6 +117,14 @@ func MustNewRegistry() featuretypes.Registry {
 			DefaultVariant: featuretypes.MustNewName("disabled"),
 			Variants:       featuretypes.NewBooleanVariants(),
 		},
+		&featuretypes.Feature{
+			Name:           FeatureUsePrometheusClickhouseV2,
+			Kind:           featuretypes.KindBoolean,
+			Stage:          featuretypes.StageExperimental,
+			Description:    "Runs PromQL queries on the clickhousev2 provider alongside the served engine result and logs any difference; serving is unaffected.",
+			DefaultVariant: featuretypes.MustNewName("disabled"),
+			Variants:       featuretypes.NewBooleanVariants(),
+		},
 	)
 	if err != nil {
 		panic(err)

pkg/modules/cloudintegration/implcloudintegration/fs/definitions/gcp/cloudsql/assets/dashboards/overview.json

@@ -0,0 +1 @@
+{}

pkg/modules/cloudintegration/implcloudintegration/fs/definitions/gcp/cloudsql/integration.json

@@ -0,0 +1,27 @@
+{
+  "id": "cloudsql",
+  "title": "GCP Cloud SQL",
+  "icon": "file://icon.svg",
+  "overview": "file://overview.md",
+  "supportedSignals": {
+    "metrics": true,
+    "logs": true
+  },
+  "dataCollected": {
+    "metrics": [],
+    "logs": []
+  },
+  "telemetryCollectionStrategy": {
+    "gcp": {}
+  },
+  "assets": {
+    "dashboards": [
+      {
+        "id": "overview",
+        "title": "GCP Cloud SQL Overview",
+        "description": "Overview of GCP Cloud SQL metrics",
+        "definition": "file://assets/dashboards/overview.json"
+      }
+    ]
+  }
+}

pkg/modules/cloudintegration/implcloudintegration/fs/definitions/gcp/cloudsql/overview.md

@@ -0,0 +1,3 @@
+### Monitor GCP Cloud SQL with SigNoz
+
+Collect key GCP Cloud SQL metrics and view them with an out of the box dashboard.

pkg/modules/cloudintegration/implcloudintegration/fs/definitions/gcp/cloudsql_postgres/integration.json

@@ -1,106 +0,0 @@
-{
-  "id": "cloudsql_postgres",
-  "title": "GCP Cloud SQL for PostgreSQL",
-  "icon": "file://icon.svg",
-  "overview": "file://overview.md",
-  "supportedSignals": {
-    "metrics": true,
-    "logs": true
-  },
-  "dataCollected": {
-    "metrics": [
-      {
-        "name": "cloudsql.googleapis.com/database/up",
-        "unit": "Count",
-        "type": "Gauge",
-        "description": ""
-      },
-      {
-        "name": "cloudsql.googleapis.com/database/cpu/utilization",
-        "unit": "Percent",
-        "type": "Gauge",
-        "description": ""
-      },
-      {
-        "name": "cloudsql.googleapis.com/database/memory/utilization",
-        "unit": "Percent",
-        "type": "Gauge",
-        "description": ""
-      },
-      {
-        "name": "cloudsql.googleapis.com/database/memory/usage",
-        "unit": "Bytes",
-        "type": "Gauge",
-        "description": ""
-      },
-      {
-        "name": "cloudsql.googleapis.com/database/disk/bytes_used",
-        "unit": "Bytes",
-        "type": "Gauge",
-        "description": ""
-      },
-      {
-        "name": "cloudsql.googleapis.com/database/postgresql/num_backends",
-        "unit": "Count",
-        "type": "Gauge",
-        "description": ""
-      },
-      {
-        "name": "cloudsql.googleapis.com/database/postgresql/num_backends_by_state",
-        "unit": "Count",
-        "type": "Gauge",
-        "description": ""
-      },
-      {
-        "name": "cloudsql.googleapis.com/database/postgresql/transaction_count",
-        "unit": "Count",
-        "type": "Gauge",
-        "description": ""
-      },
-      {
-        "name": "cloudsql.googleapis.com/database/postgresql/deadlock_count",
-        "unit": "Count",
-        "type": "Gauge",
-        "description": ""
-      },
-      {
-        "name": "cloudsql.googleapis.com/database/postgresql/vacuum/oldest_transaction_age",
-        "unit": "Count",
-        "type": "Gauge",
-        "description": ""
-      },
-      {
-        "name": "cloudsql.googleapis.com/database/postgresql/insights/aggregate/execution_time",
-        "unit": "Microseconds",
-        "type": "Gauge",
-        "description": ""
-      },
-      {
-        "name": "cloudsql.googleapis.com/database/postgresql/insights/perquery/execution_time",
-        "unit": "Microseconds",
-        "type": "Gauge",
-        "description": ""
-      },
-      {
-        "name": "cloudsql.googleapis.com/database/postgresql/replication/replica_byte_lag",
-        "unit": "Bytes",
-        "type": "Gauge",
-        "description": ""
-      }
-    ],
-    "logs": []
-  },
-  "telemetryCollectionStrategy": {
-    "gcp": {}
-  },
-  "assets": {
-    "dashboards": [
-      {
-        "id": "overview",
-        "title": "GCP Cloud SQL for PostgreSQL Overview",
-        "description": "Overview of GCP Cloud SQL for PostgreSQL metrics",
-        "definition": "file://assets/dashboards/overview.json"
-      }
-    ]
-  }
-}

pkg/modules/cloudintegration/implcloudintegration/fs/definitions/gcp/cloudsql_postgres/overview.md

@@ -1,3 +0,0 @@
-### Monitor GCP Cloud SQL for PostgreSQL with SigNoz
-
-Collect key GCP Cloud SQL for PostgreSQL metrics and view them with an out of the box dashboard.

pkg/modules/inframonitoring/implinframonitoring/checks_constants.go

@@ -16,11 +16,11 @@ const (
 
 // Documentation links — one per component. User-facing; emitted on missing-entries.
 const (
-	docLinkHostMetricsReceiver        = "https://signoz.io/docs/infrastructure-monitoring/hostmetrics/#configure-the-hostmetrics-receiver"
-	docLinkKubeletStatsReceiver       = "https://signoz.io/docs/infrastructure-monitoring/k8s-metrics/#2-configure-the-kubelet-stats-receiver"
-	docLinkK8sClusterReceiver         = "https://signoz.io/docs/infrastructure-monitoring/k8s-metrics/#1-configure-the-k8s-cluster-receiver"
-	docLinkResourceDetectionProcessor = "https://signoz.io/docs/infrastructure-monitoring/hostmetrics/#configure-the-processors"
-	docLinkK8sAttributesProcessor     = "https://signoz.io/docs/infrastructure-monitoring/k8s-metrics/#3-enable-kubernetes-metadata"
+	docLinkHostMetricsReceiver        = "https://signoz.io/docs/infrastructure-monitoring/user-guides/hostmetrics/#configure-the-hostmetrics-receiver"
+	docLinkKubeletStatsReceiver       = "https://signoz.io/docs/infrastructure-monitoring/user-guides/k8s-metrics/#setup-kubelet-stats-receiver"
+	docLinkK8sClusterReceiver         = "https://signoz.io/docs/infrastructure-monitoring/user-guides/k8s-metrics/#setup-k8s-cluster-receiver"
+	docLinkResourceDetectionProcessor = "https://signoz.io/docs/infrastructure-monitoring/user-guides/hostmetrics/#configure-the-resourcedetection-processor"
+	docLinkK8sAttributesProcessor     = "https://signoz.io/docs/infrastructure-monitoring/user-guides/k8s-metrics/#3-setup-k8sattributesprocessor-to-enable-kubernetes-metadata"
 )
 
 var (

pkg/prometheus/clickhouseprometheusv2/capture.go

@@ -0,0 +1,89 @@
+package clickhouseprometheusv2
+
+import (
+	"context"
+	"sync"
+
+	"github.com/SigNoz/signoz/pkg/prometheus"
+	"github.com/prometheus/prometheus/model/labels"
+	"github.com/prometheus/prometheus/storage"
+	"github.com/prometheus/prometheus/util/annotations"
+)
+
+// statementRecorder collects the statements a PromQL evaluation would run.
+// Safe for concurrent use: the engine may Select selectors concurrently.
+type statementRecorder struct {
+	mu         sync.Mutex
+	statements []prometheus.CapturedStatement
+}
+
+func (r *statementRecorder) record(query string, args []any) {
+	r.mu.Lock()
+	defer r.mu.Unlock()
+	r.statements = append(r.statements, prometheus.CapturedStatement{Query: query, Args: args})
+}
+
+func (r *statementRecorder) Statements() []prometheus.CapturedStatement {
+	r.mu.Lock()
+	defer r.mu.Unlock()
+	out := make([]prometheus.CapturedStatement, len(r.statements))
+	copy(out, r.statements)
+	return out
+}
+
+type captureQueryable struct {
+	client   *client
+	recorder *statementRecorder
+}
+
+func (c *captureQueryable) Querier(mint, maxt int64) (storage.Querier, error) {
+	return &captureQuerier{
+		querier:  querier{mint: mint, maxt: maxt, client: c.client},
+		recorder: c.recorder,
+	}, nil
+}
+
+// captureQuerier builds the same SQL as the live querier but records it and
+// returns no data. The fingerprint filter always takes the subquery form:
+// without executing the series lookup, the inline literal set is unknown.
+type captureQuerier struct {
+	querier
+	recorder *statementRecorder
+}
+
+func (c *captureQuerier) Select(ctx context.Context, _ bool, hints *storage.SelectHints, matchers ...*labels.Matcher) storage.SeriesSet {
+	if rawQuery, ok := rawSQLQuery(matchers); ok {
+		c.recorder.record(rawQuery, nil)
+		return storage.EmptySeriesSet()
+	}
+
+	start, end := c.window(hints)
+
+	samplesQuery, args, err := buildSamplesQuery(start, end, metricNamesFromMatchers(matchers), nil, matchers, c.lastSamplePerStepFor(ctx, hints))
+	if err != nil {
+		return storage.ErrSeriesSet(err)
+	}
+	c.recorder.record(samplesQuery, args)
+
+	return storage.EmptySeriesSet()
+}
+
+func (c *captureQuerier) LabelValues(context.Context, string, *storage.LabelHints, ...*labels.Matcher) ([]string, annotations.Annotations, error) {
+	return nil, nil, nil
+}
+
+func (c *captureQuerier) LabelNames(context.Context, *storage.LabelHints, ...*labels.Matcher) ([]string, annotations.Annotations, error) {
+	return nil, nil, nil
+}
+
+// metricNamesFromMatchers extracts the statically known metric name, if any.
+// The live path derives names from the matched series; the capture path has
+// no execution results, so only a __name__ equality contributes.
+func metricNamesFromMatchers(matchers []*labels.Matcher) []string {
+	for _, m := range matchers {
+		if m.Name == metricNameLabel && m.Type == labels.MatchEqual && m.Value != "" {
+			return []string{m.Value}
+		}
+	}
+	return nil
+}

pkg/prometheus/clickhouseprometheusv2/client.go

@@ -0,0 +1,282 @@
+package clickhouseprometheusv2
+
+import (
+	"context"
+	"database/sql"
+	"encoding/json"
+	"fmt"
+	"log/slog"
+	"math"
+	"slices"
+
+	"github.com/SigNoz/signoz/pkg/errors"
+	"github.com/SigNoz/signoz/pkg/factory"
+	"github.com/SigNoz/signoz/pkg/prometheus"
+	"github.com/SigNoz/signoz/pkg/telemetrystore"
+	"github.com/SigNoz/signoz/pkg/types/ctxtypes"
+	"github.com/SigNoz/signoz/pkg/types/instrumentationtypes"
+	"github.com/SigNoz/signoz/pkg/types/telemetrytypes"
+	"github.com/prometheus/prometheus/model/labels"
+	promValue "github.com/prometheus/prometheus/model/value"
+)
+
+// seriesLookup is a series-lookup result: matched fingerprints with their
+// labels, and the distinct metric names seen on them.
+type seriesLookup struct {
+	fingerprints map[uint64]labels.Labels
+	metricNames  []string
+}
+
+// client executes the series, samples and raw queries against ClickHouse.
+type client struct {
+	settings       factory.ScopedProviderSettings
+	telemetryStore telemetrystore.TelemetryStore
+	cfg            prometheus.ClickhouseV2Config
+	lookbackMs     int64
+}
+
+func newClient(settings factory.ScopedProviderSettings, telemetryStore telemetrystore.TelemetryStore, cfg prometheus.Config) *client {
+	lookback := cfg.LookbackDelta
+	if lookback <= 0 {
+		// Mirror the engine: promql defaults an unset lookback to 5m.
+		lookback = defaultLookbackDelta
+	}
+	return &client{
+		settings:       settings,
+		telemetryStore: telemetryStore,
+		cfg:            cfg.ClickhouseV2,
+		lookbackMs:     lookback.Milliseconds(),
+	}
+}
+
+func (c *client) withContext(ctx context.Context, functionName string) context.Context {
+	return ctxtypes.NewContextWithCommentVals(ctx, map[string]string{
+		instrumentationtypes.TelemetrySignal:  telemetrytypes.SignalMetrics.StringValue(),
+		instrumentationtypes.CodeNamespace:    "clickhouse-prometheus-v2",
+		instrumentationtypes.CodeFunctionName: functionName,
+	})
+}
+
+// selectSeries runs the series lookup for the given matchers and window.
+func (c *client) selectSeries(ctx context.Context, query string, args []any) (*seriesLookup, error) {
+	ctx = c.withContext(ctx, "selectSeries")
+	rows, err := c.telemetryStore.ClickhouseDB().Query(ctx, query, args...)
+	if err != nil {
+		return nil, err
+	}
+	defer rows.Close()
+
+	lookup := &seriesLookup{fingerprints: make(map[uint64]labels.Labels)}
+	names := make(map[string]struct{})
+
+	var fingerprint uint64
+	var labelsJSON string
+	for rows.Next() {
+		if err := rows.Scan(&fingerprint, &labelsJSON); err != nil {
+			return nil, err
+		}
+		lset, err := unmarshalLabels(labelsJSON)
+		if err != nil {
+			return nil, err
+		}
+		lookup.fingerprints[fingerprint] = lset
+		if name := lset.Get(metricNameLabel); name != "" {
+			names[name] = struct{}{}
+		}
+		if c.cfg.MaxFetchedSeries > 0 && len(lookup.fingerprints) > c.cfg.MaxFetchedSeries {
+			return nil, errors.NewInvalidInputf(
+				errors.CodeInvalidInput,
+				"promql selector matched more than %d series; narrow the label matchers or raise prometheus::clickhousev2::max_fetched_series",
+				c.cfg.MaxFetchedSeries,
+			)
+		}
+	}
+	if err := rows.Err(); err != nil {
+		return nil, err
+	}
+
+	for name := range names {
+		lookup.metricNames = append(lookup.metricNames, name)
+	}
+	slices.Sort(lookup.metricNames)
+
+	return lookup, nil
+}
+
+// unmarshalLabels parses the labels JSON column. Unlike v1, the fingerprint
+// is not injected as a synthetic label (it would take part in `without (...)`
+// grouping and vector matching) and empty-valued labels are dropped: an empty
+// label value means "label absent" in Prometheus, and upstream never produces
+// such labels, but stored attribute JSON can carry them.
+func unmarshalLabels(s string) (labels.Labels, error) {
+	m := make(map[string]string)
+	if err := json.Unmarshal([]byte(s), &m); err != nil {
+		return labels.EmptyLabels(), err
+	}
+	builder := labels.NewScratchBuilder(len(m))
+	for k, v := range m {
+		if v == "" {
+			continue
+		}
+		builder.Add(k, v)
+	}
+	builder.Sort()
+	return builder.Labels(), nil
+}
+
+// selectSamples executes a samples query (raw or last-sample-per-step; both
+// produce the same column shape) and assembles the per-series sample slices.
+// Rows arrive ordered by (fingerprint, unix_milli). Rows whose fingerprint
+// is missing from the lookup are skipped (possible in the subquery filter
+// mode, where the fingerprint filter re-runs after the lookup and can see
+// series born in between). Stale flags map to the engine's StaleNaN.
+// Duplicate timestamps pass through as stored: upstream Prometheus cannot
+// produce them (its TSDB rejects them at ingest), our ingest can under
+// at-least-once retries, and v1 feeds them to the engine as-is —
+// deduplicating here would make this provider silently disagree with both
+// v1 and the transpiled statements over the same dirty data. Uniqueness
+// belongs to the ingest layer.
+func (c *client) selectSamples(ctx context.Context, query string, args []any, lookup *seriesLookup) ([]*series, error) {
+	ctx = c.withContext(ctx, "selectSamples")
+	rows, err := c.telemetryStore.ClickhouseDB().Query(ctx, query, args...)
+	if err != nil {
+		return nil, err
+	}
+	defer rows.Close()
+
+	var (
+		result       []*series
+		current      *series
+		fingerprint  uint64
+		prevFp       uint64
+		timestampMs  int64
+		val          float64
+		flags        uint32
+		first        = true
+		haveCurrent  bool
+		staleMarker  = math.Float64frombits(promValue.StaleNaN)
+		maxSamples   = c.cfg.MaxFetchedSamples
+		fetched      int64
+		unknownCount int
+	)
+
+	for rows.Next() {
+		if err := rows.Scan(&fingerprint, &timestampMs, &val, &flags); err != nil {
+			return nil, err
+		}
+
+		fetched++
+		if maxSamples > 0 && fetched > maxSamples {
+			return nil, errors.NewInvalidInputf(
+				errors.CodeInvalidInput,
+				"promql query would fetch more than %d samples; narrow the selector or time range, or raise prometheus::clickhousev2::max_fetched_samples",
+				maxSamples,
+			)
+		}
+
+		if first || fingerprint != prevFp {
+			first = false
+			prevFp = fingerprint
+			lset, ok := lookup.fingerprints[fingerprint]
+			if !ok {
+				unknownCount++
+				haveCurrent = false
+				continue
+			}
+			current = &series{lset: lset}
+			result = append(result, current)
+			haveCurrent = true
+		}
+		if !haveCurrent {
+			// Remaining rows of a fingerprint missing from the lookup.
+			continue
+		}
+
+		if flags&1 == 1 {
+			val = staleMarker
+		}
+		current.ts = append(current.ts, timestampMs)
+		current.vs = append(current.vs, val)
+	}
+	if err := rows.Err(); err != nil {
+		return nil, err
+	}
+
+	if unknownCount > 0 {
+		c.settings.Logger().DebugContext(ctx, "skipped samples of fingerprints missing from series lookup",
+			slog.Int("unknown_fingerprints", unknownCount))
+	}
+
+	return result, nil
+}
+
+// queryRaw supports the {job="rawsql", query="..."} escape hatch: the value of
+// the query matcher runs as-is, each row becoming a single-sample series
+// stamped at the query end. Column "value" is the sample value; every other
+// column is a label.
+func (c *client) queryRaw(ctx context.Context, query string, ts int64) ([]*series, error) {
+	ctx = c.withContext(ctx, "queryRaw")
+	rows, err := c.telemetryStore.ClickhouseDB().Query(ctx, query)
+	if err != nil {
+		return nil, err
+	}
+	defer rows.Close()
+
+	columns := rows.Columns()
+	targets := make([]any, len(columns))
+	for i := range targets {
+		targets[i] = new(scanner)
+	}
+
+	var result []*series
+	for rows.Next() {
+		if err := rows.Scan(targets...); err != nil {
+			return nil, err
+		}
+		builder := labels.NewScratchBuilder(len(columns))
+		var val float64
+		for i, col := range columns {
+			v := targets[i].(*scanner)
+			if col == "value" {
+				val = v.f
+				continue
+			}
+			builder.Add(col, v.s)
+		}
+		builder.Sort()
+		result = append(result, &series{
+			lset: builder.Labels(),
+			ts:   []int64{ts},
+			vs:   []float64{val},
+		})
+	}
+	if err := rows.Err(); err != nil {
+		return nil, err
+	}
+	return result, nil
+}
+
+var _ sql.Scanner = (*scanner)(nil)
+
+type scanner struct {
+	f float64
+	s string
+}
+
+func (s *scanner) Scan(val any) error {
+	s.f = 0
+	s.s = ""
+
+	s.s = fmt.Sprintf("%v", val)
+	switch val := val.(type) {
+	case int64:
+		s.f = float64(val)
+	case uint64:
+		s.f = float64(val)
+	case float64:
+		s.f = val
+	case []byte:
+		s.s = string(val)
+	}
+	return nil
+}

pkg/prometheus/clickhouseprometheusv2/doc.go

@@ -0,0 +1,334 @@
+// Package clickhouseprometheusv2 is the second-generation ClickHouse-backed
+// Prometheus provider. It exists because the v1 provider fetches every raw
+// sample of a query's union window through the remote-read protobuf layer
+// and hands it to the engine — the cost is a function of ingested data, not
+// of the question asked, which is how a dashboard of PromQL panels takes an
+// instance down.
+//
+// Every query runs in one of two ways, decided per query:
+//
+//   - Transpiled: the query is evaluated entirely inside ClickHouse and only
+//     final (or near-final) per-group grid arrays come back, built on the
+//     timeSeries*ToGrid aggregate functions (the supported ClickHouse floor
+//     is >= 25.6, so they are assumed available).
+//   - Engine: the stock promql engine evaluates over this package's native
+//     storage.Querier. This is the path for everything not transpilable.
+//
+// Correctness is the constraint that shaped both paths: a PromQL result that
+// differs from upstream Prometheus is a lost user, so anything that cannot
+// reproduce engine semantics exactly falls back rather than approximate.
+// The rest of this comment is the PromQL -> SQL story, because that mapping
+// is where correctness is won or lost.
+//
+// # The evaluation model the SQL must reproduce
+//
+// A PromQL range query is an instant query evaluated at every grid point
+// t_i = start + i*step, i = 0..(end-start)/step. At each t_i:
+//
+//   - an instant selector resolves to the latest sample in the left-open
+//     lookback window (t_i - lookback, t_i], and to nothing when that latest
+//     sample is a stale marker — even if older real samples sit inside the
+//     window;
+//   - a range selector [r] collects every sample in (t_i - r, t_i], stale
+//     markers excluded;
+//   - offset d shifts both windows to (t_i - d - w, t_i - d].
+//
+// The transpilation invariant follows from this: every transpiled construct
+// produces, per output series, one array with exactly one slot per grid
+// point — slot i holds the value at t_i, NULL means absent. This is what
+// makes composition correct, not just convenient: the engine evaluates
+// these operators independently per t_i, so any representation that gets
+// every slot right gets the whole query right, and spatial aggregation over
+// arrays is sound because it combines values that belong to the same t_i by
+// construction. Slot index i maps back to t_i = start + i*step at scan time
+// (toMatrix). Everything below is about filling those slots with exactly
+// the numbers the engine would compute — and each equivalence was validated
+// against the vendored engine on live data before its shape entered the
+// allowlist; anything unproven stays on the engine path.
+//
+// # Classification: finding what a statement can answer
+//
+// classify walks the parsed AST looking for "core units" — maximal subtrees
+// of the shape
+//
+//	[agg by/without (...)] [fn(] selector[range] [offset d] [)] [op scalar]...
+//
+// classifyCore peels that chain from the outside in: an optional
+// sum/min/max/avg/count aggregation, then one of the allowlisted functions
+// or a bare instant selector, then the selector with its offset; on the way
+// out it accumulates number-literal arithmetic, comparisons (including
+// bool) and unary minus into a scalar-op pipeline. A node qualifies only if
+// its type, arguments and children are in the proven set — an allowlist, so
+// an overlooked construct becomes a fallback instead of a wrong number.
+//
+// Three unit kinds come out of this, each with its own SQL form:
+// unitRange (rate, irate, increase, delta, idelta over a range selector),
+// unitInstant (instant vector selection, bare or comparison-filtered) and
+// unitOverTime (avg/min/max/sum/count/last _over_time).
+//
+// If the entire tree is one unit, the plan is "full": the statement's rows
+// are the query result. Otherwise every maximal unit is cut out and replaced
+// in the expression with a synthetic selector __signoz_transpiled_N__, and
+// the rewritten expression runs in the engine over the units' materialized
+// results ("hybrid") — histogram_quantile, topk, or/and/unless and vector
+// matching keep exact engine semantics while their expensive inputs were
+// aggregated server-side.
+//
+// Classification refuses when exact semantics cannot be guaranteed
+// server-side: the @ modifier anywhere and default-resolution subqueries
+// (their resolution is a server runtime setting the transpiler cannot see);
+// steps or ranges that are not whole seconds (the grid functions take
+// whole-second parameters); grouping by or matching on __name__ in hybrid
+// plans (the synthetic name would leak into results); name-keeping units —
+// bare/comparison instant selectors and last_over_time keep their real
+// __name__ (keepsName), which substitution would replace, so they transpile
+// only as full plans; and every function outside the allowlist (changes,
+// resets, quantile_over_time, absent, native-histogram functions, ...).
+//
+// Units inside a fixed-resolution subquery evaluate on the subquery's own
+// grid instead of the query grid: epoch-aligned multiples of the resolution
+// strictly after outerStart - offset - range, ending at outer end - offset —
+// the exact derivation the engine uses, because a grid shifted by one step
+// changes which samples every window sees.
+//
+// # From one unit to one statement
+//
+// buildUnitSQL renders each unit as a single statement. For
+// sum by (pod) (rate(m{job="api"}[5m])) the skeleton is:
+//
+//	SELECT gkey, sumForEach(grid) AS grid FROM (
+//	    SELECT series.gkey AS gkey,
+//	           timeSeriesRateToGrid(<start>, <end>, <step>, <range>)(fromUnixTimestamp64Milli(unix_milli), value) AS grid
+//	    FROM signoz_metrics.distributed_samples_v4 AS points
+//	    INNER JOIN (
+//	        SELECT fingerprint, <group key expr> AS gkey
+//	        FROM signoz_metrics.time_series_v4
+//	        WHERE <series predicates>
+//	        GROUP BY fingerprint, gkey
+//	    ) AS series ON points.fingerprint = series.fingerprint
+//	    WHERE metric_name = ? AND temporality IN ['Cumulative', 'Unspecified']
+//	      AND points.fingerprint IN (<matched fingerprints>)
+//	      AND unix_milli > <start - range> AND unix_milli <= <end>
+//	      AND bitAnd(flags, 1) = 0
+//	    GROUP BY points.fingerprint, series.gkey
+//	) GROUP BY gkey
+//	SETTINGS allow_experimental_ts_to_grid_aggregate_function = 1
+//
+// Reading it inside out:
+//
+// The time window is the selector's semantics verbatim: strict > on the
+// lower bound and <= on the upper is the left-open (t - w, t] rule, with the
+// whole window shifted by the offset. bitAnd(flags, 1) = 0 drops stale
+// markers, which PromQL excludes from range vectors.
+//
+// The inner GROUP BY computes one grid array per series.
+// timeSeriesRateToGrid(start, end, step, range) is a parametric aggregate:
+// fed (timestamp, value) pairs it produces Array(Nullable(Float64)) with one
+// slot per grid point. Correct because it implements the engine's
+// extrapolatedRate decision for decision — counter resets, the zero-point
+// clamp, the extrapolation thresholds, the >= 2 samples rule, the left-open
+// window — verified by feeding identical samples to both and comparing
+// slot for slot: the only difference ever observed is the last bit
+// (ClickHouse's C++ and Go round the same formula differently), which is
+// the floating-point floor, not a semantic gap. irate/delta/idelta map to
+// their own timeSeries*ToGrid functions with the same verification;
+// increase has no function of its own and is emitted as
+// arrayMap(x -> x * <range seconds>, <rate expr>), exact by definition —
+// extrapolatedRate computes the same extrapolated delta for both and
+// divides by the range only when isRate, so multiplying it back is the
+// identity, not an approximation. The grid parameters are rendered as
+// literals, not bound args — they are aggregate-function parameters — and
+// the experimental gate rides as a SETTINGS clause on the statement itself
+// so telemetrystore hooks cannot clobber it.
+//
+// The join annotates each series with its group key: toJSONString of the
+// sorted [label, value] pairs the unit projects, extracted from the stored
+// labels JSON. by keeps the listed labels, without excludes them plus
+// __name__, no aggregation keeps everything minus __name__ unless the unit
+// keeps its name — the engine's name-dropping rules. Correct as a grouping
+// key because the pairs are sorted and empty values are filtered: key
+// equality is then exactly label-set equality on the projection —
+// Prometheus treats an empty label value as the label being absent, and
+// stored attribute JSON can carry empties that must not split groups — and
+// the same canonical string parses back into the output label set
+// (labelsFromGroupKey).
+//
+// The outer GROUP BY is the spatial aggregation: sum/min/max/avg/count
+// by/without become the -ForEach combinators. Element-wise aggregation over
+// grid arrays is the engine's per-t_i aggregation, because slot i of every
+// input array refers to the same t_i; the combinators skip NULLs, which is
+// the engine aggregating only the series present at t_i, and an index where
+// every series is absent stays NULL. Two edges need explicit handling:
+// countForEach wraps in a mapping of 0 back to NULL, because a count over
+// an all-absent index is an absent point, not 0; and a unit without
+// aggregation still passes through maxForEach — the identity for the common
+// one-fingerprint group, and a deterministic NULL-skipping merge when a
+// regex __name__ selector collapses distinct metrics onto one projected
+// label set. One caveat is inherent: summation order over series differs
+// from the engine's, so spatial aggregates can differ in the last ULP —
+// float addition is not associative; no ordering reproduces the engine's
+// bit-exactly from inside a GROUP BY.
+//
+// # Instant selectors: staleness needs two aggregates
+//
+// unitInstant uses window = lookback and must reproduce the shadowing rule:
+// the point is absent when the latest in-window sample is a stale marker.
+// timeSeriesLastToGrid alone cannot express that — skipping stale rows in
+// WHERE would resurrect the older real sample the marker was written to
+// bury. So stale rows stay in the scan for this kind only, and the grid
+// expression compares three aggregates per slot:
+//
+//	arrayMap((tall, tok, vok) -> if(tall IS NULL OR tok IS NULL OR tall != tok, NULL, vok),
+//	         timeSeriesLastToGrid(...)(ts, toFloat64(unix_milli)),                           -- last sample overall
+//	         timeSeriesLastToGridIf(...)(ts, toFloat64(unix_milli), bitAnd(flags, 1) = 0),   -- last non-stale, its timestamp
+//	         timeSeriesLastToGridIf(...)(ts, value, bitAnd(flags, 1) = 0))                   -- last non-stale, its value
+//
+// Correct by cases on a slot's window. No samples at all: both timestamp
+// aggregates are NULL, the slot is NULL — absent, as the engine says. Latest
+// sample non-stale: it is the latest overall and the latest non-stale, the
+// timestamps agree, the slot takes its value — the engine's pick. Latest
+// sample stale: the last-overall timestamp is the marker's, the
+// last-non-stale timestamp is older (or NULL when only markers are in
+// window), they disagree, the slot is NULL — the marker shadows, exactly
+// the engine's rule. Timestamps are unique per series (ingest dedups), so
+// timestamp equality identifies "the same sample" without ambiguity. The
+// -If combinator's applicability to these experimental aggregates was
+// probed before being trusted, not assumed.
+//
+// # Windowed *_over_time: fan-out instead of a grid function
+//
+// avg/min/max/sum/count _over_time aggregate every raw sample in the window,
+// and no timeSeries*ToGrid function computes them. (last_over_time is the
+// exception: the last sample of a range vector — stale markers excluded from
+// range vectors by PromQL, excluded here in WHERE — is exactly
+// timeSeriesLastToGrid.) Instead, each sample is fanned out to every grid
+// index whose window contains it:
+//
+//	ARRAY JOIN range(toUInt64(greatest(0, intDiv(unix_milli - <start> + <step> - 1, <step>))),
+//	                 toUInt64(least(<lastIdx>, intDiv(unix_milli + <range> - 1 - <start>, <step>)) + 1)) AS k
+//
+// Correct because the bounds solve the window condition for k. A sample at
+// ts contributes to slot k iff t_k - range < ts <= t_k. The right side
+// gives t_k >= ts, so the first index is ceil((ts - start)/step) — a sample
+// at exactly t_k belongs to k, the window is right-closed. The left side
+// gives t_k < ts + range, and with millisecond-integer timestamps that is
+// t_k <= ts + range - 1, so the last index is
+// floor((ts + range - 1 - start)/step) — a sample at exactly t_k - range is
+// excluded, the window is left-open. Clamped to the grid, the fan-out
+// therefore lands each sample in exactly the slots whose windows contain
+// it, and GROUP BY (fingerprint, k) with the plain aggregate (avg(value),
+// min(value), ...) computes per slot over precisely the engine's sample
+// multiset — the same numbers, since avg/min/max/sum/count are
+// order-insensitive on a given multiset (sum/avg up to summation order, the
+// float caveat above). A second level assembles the positional array with
+// groupArray + indexOf, mapping missing indices to NULL — groupArrayInsertAt
+// would coerce NULL defaults to 0, which is a value, not absence. The
+// group-key join happens at the initiator here, over rows already reduced
+// to per-(series, index); see the sharding section for why that costs
+// nothing.
+//
+// # Scalar ops, full plans, hybrid plans
+//
+// The scalar-op pipeline applies in Go to the returned arrays
+// (applyScalarOps), slot by slot: arithmetic operators compute, comparisons
+// filter (the slot keeps the vector-side value or becomes NULL) or return
+// 0/1 under bool. Correct trivially: it is the same float64 operation the
+// engine would apply to the same slot value, in the same operator order the
+// AST dictates — running it in Go instead of another SQL layer changes
+// where, not what.
+//
+// A full plan's arrays map straight to the result matrix. A hybrid plan
+// materializes each unit's arrays as synthetic series under its
+// __signoz_transpiled_N__ name and evaluates the rewritten expression over
+// a storage that serves synthetic names from memory and everything else
+// live. Substitution is sound because a unit's output is a plain instant
+// vector to the engine — same values at same timestamps under a different
+// name, and the name cannot matter: plans that group by or match on
+// __name__ were refused at classification, and name-keeping units are never
+// substituted. One subtlety makes it exact: stale markers are written at
+// absent grid points, because the engine's lookback would otherwise
+// resurrect a point from up to lookback earlier — the marker encodes
+// "absent here" the way the engine itself encodes it. Units evaluate
+// concurrently; each is one series lookup plus one grid statement. A step
+// of 0 is an instant query: a single evaluation at end.
+//
+// # Series lookup
+//
+// Both paths resolve matchers the same way, once per selector
+// (selectSeries): __name__ matchers translate to the metric_name column —
+// all four matcher types; the v1 client silently returned nothing for regex
+// metric names — and every other matcher to a JSONExtractString condition on
+// the labels column (applySeriesConditions). Regexes are anchored before
+// they reach match(): PromQL matchers match the whole value, ClickHouse
+// match() searches for a substring, and without anchoring =~"api" would
+// also select "x-api-y". An equality matcher against "" matches series
+// without the label, mirroring PromQL, because JSONExtractString returns ""
+// for missing keys. The series tables hold one row per (fingerprint, bucket)
+// at 1h/6h/1d/1w granularities; timeSeriesTableFor picks the table whose
+// bucket fits the window and rounds the window start down to the bucket
+// boundary. The resulting label sets drop what v1 leaked into results: the
+// synthetic fingerprint label (it would take part in without() grouping and
+// vector matching) and empty-valued labels. MaxFetchedSeries fails the
+// lookup with a typed invalid-input error past the ceiling — v1's behavior
+// for an oversized selector was to buffer everything and OOM, and a 4xx the
+// user can narrow beats a dead process serving nobody.
+//
+// # The engine path
+//
+// Queries that do not transpile run in the stock engine over this package's
+// storage.Querier, which is still not the v1 path. Samples are fetched per
+// selector using the engine's per-selector hints, not the query-wide union
+// window, so foo / foo offset 1d reads two narrow windows instead of the
+// widest one twice. Instant selectors of subquery-free queries fetch only
+// the last sample per step bucket (lastSamplePerStep): buckets anchor at the
+// selector's first evaluation timestamp — recovered from the hints as
+// hints.Start + lookback - 1ms, the inverse of how the engine derives
+// hints.Start — so bucket boundaries coincide with evaluation timestamps and
+// a non-final sample of a bucket can never be the latest sample in
+// (t - lookback, t] for any grid t. Real timestamps are preserved, so the
+// engine's own lookback and staleness handling stay exact. Range selectors
+// always fetch raw — every sample feeds the range function — and the
+// subquery-free proof travels in the context as prometheus.QueryTraits,
+// because subquery selectors evaluate at the subquery's step while the
+// hints carry the top-level step. Row assembly counts rows against
+// MaxFetchedSamples while scanning, keeps the first of consecutive equal
+// timestamps, maps stale flags to the engine's StaleNaN, and merges series
+// with identical label sets (sortAndMerge) — the engine assumes storages
+// never emit duplicates. A {job="rawsql", query="..."} selector bypasses all
+// of this and runs the query matcher's value verbatim.
+//
+// # Sharding
+//
+// samples_v4 and time_series_v4 (and all their rollups) shard on the same
+// key — cityHash64(env, temporality, metric_name, fingerprint) — so a
+// series' samples and catalog rows live on the same shard. The transpiled
+// statement above exploits that: the distributed samples table at the
+// top-level FROM makes ClickHouse rewrite the whole inner query per shard,
+// where the join against the shard-local series table and the per-series
+// grid aggregation run next to the data; the initiator only merges
+// aggregate states and applies the spatial -ForEach step. Same layout as
+// the telemetrymetrics statement builder. Fingerprint filters follow suit:
+// matched sets inline as sorted literals up to inlineFingerprintsLimit
+// (literals engage the samples primary key; sorting keeps statements
+// deterministic), beyond it the group-key join alone restricts — a
+// semi-join on the same predicates would only rescan the series table —
+// except the windowed *_over_time fan-out, which has no join and keeps a
+// shard-local IN subquery rather than expand every series of the metric.
+// The engine path's over-limit filter is the same shard-local subquery, not
+// a GLOBAL broadcast of the matched set. The temporality filter on every
+// samples statement is a semantic no-op — the matched fingerprints already
+// come from those temporalities — that engages the leading samples
+// primary-key column. Delta-temporality series stay invisible to PromQL
+// here exactly as they are in v1: the rollout gate is parity with v1, and
+// making Delta visible is its own change with its own semantics to design —
+// a Delta stream fed to rate() as-if-cumulative would be wrong, not just
+// new.
+//
+// # Observability
+//
+// Every statement carries a log_comment with
+// code.namespace=clickhouse-prometheus-v2 and code.function.name naming the
+// call site (selectSeries, selectSamples, transpiledUnit, LabelValues,
+// LabelNames), so this provider's work is attributable in system.query_log
+// without guessing from query text.
+package clickhouseprometheusv2

pkg/prometheus/clickhouseprometheusv2/last_sample_per_step_test.go

@@ -0,0 +1,191 @@
+package clickhouseprometheusv2
+
+import (
+	"fmt"
+	"math/rand"
+	"sort"
+	"testing"
+
+	"github.com/stretchr/testify/require"
+)
+
+// The lastSamplePerStep correctness argument, executed: for instant selectors, keeping
+// only the last sample of every step bucket (bucket 0 = (start, firstEval],
+// bucket i = (firstEval+(i-1)·step, firstEval+i·step]) yields exactly the
+// same instant-vector selections as the raw samples, for every evaluation
+// timestamp on the grid. The engine picks the latest sample in
+// (t-lookback, t] per evaluation timestamp t and treats a stale marker as
+// absent; both behaviors are emulated here directly.
+
+type tsample struct {
+	ts    int64
+	value float64
+	stale bool
+}
+
+// engineSelect emulates the engine's instant-selector resolution at
+// evaluation timestamp t over samples ordered by timestamp: the latest sample
+// in (t-lookback, t], absent when none or when it is a stale marker.
+func engineSelect(samples []tsample, t, lookbackMs int64) (tsample, bool) {
+	var picked tsample
+	found := false
+	for _, s := range samples {
+		if s.ts > t-lookbackMs && s.ts <= t {
+			picked = s
+			found = true
+		}
+	}
+	if !found || picked.stale {
+		return tsample{}, false
+	}
+	return picked, true
+}
+
+// lastPerStep emulates the last-sample-per-step samples query: group samples into buckets and
+// keep only the last sample of each (ties keep either; ClickHouse argMax over
+// equal keys is unspecified, so generated timestamps are unique).
+func lastPerStep(samples []tsample, firstEvalMs, stepMs int64) []tsample {
+	last := make(map[int64]tsample)
+	for _, s := range samples {
+		var bucket int64
+		if stepMs > 0 && s.ts > firstEvalMs {
+			bucket = (s.ts-firstEvalMs-1)/stepMs + 1
+		}
+		if cur, ok := last[bucket]; !ok || s.ts > cur.ts {
+			last[bucket] = s
+		}
+	}
+	out := make([]tsample, 0, len(last))
+	for _, s := range last {
+		out = append(out, s)
+	}
+	sort.Slice(out, func(i, j int) bool { return out[i].ts < out[j].ts })
+	return out
+}
+
+func TestLastSamplePerStepEquivalence(t *testing.T) {
+	rng := rand.New(rand.NewSource(42))
+
+	for caseIdx := 0; caseIdx < 2000; caseIdx++ {
+		// Random query shape. Units are milliseconds but kept small so bucket
+		// boundaries are hit often.
+		stepMs := []int64{1, 2, 5, 7, 30, 60}[rng.Intn(6)]
+		lookbackMs := []int64{1, 3, 5, 10, 45}[rng.Intn(5)]
+		queryStart := int64(1000)
+		numSteps := rng.Int63n(20)
+		queryEnd := queryStart + numSteps*stepMs + rng.Int63n(stepMs) // grid may not divide the range
+
+		// Engine-derived selector window for instant selectors:
+		// hints.Start = firstEval - (lookback - 1), hints.End = queryEnd.
+		hintsStart := queryStart - (lookbackMs - 1)
+		hintsEnd := queryEnd
+		firstEval := hintsStart + lookbackMs - 1
+		require.Equal(t, queryStart, firstEval)
+
+		// Random samples inside the fetch window [hints.Start, hints.End],
+		// with unique timestamps and occasional stale markers. The sample
+		// count is capped by the window size: timestamps are unique.
+		windowSize := hintsEnd - hintsStart + 1
+		numSamples := rng.Int63n(40)
+		if numSamples > windowSize {
+			numSamples = windowSize
+		}
+		seen := make(map[int64]bool)
+		var samples []tsample
+		for int64(len(samples)) < numSamples {
+			ts := hintsStart + rng.Int63n(windowSize)
+			if seen[ts] {
+				continue
+			}
+			seen[ts] = true
+			samples = append(samples, tsample{ts: ts, value: rng.Float64(), stale: rng.Intn(8) == 0})
+		}
+		sort.Slice(samples, func(i, j int) bool { return samples[i].ts < samples[j].ts })
+
+		reduced := lastPerStep(samples, firstEval, stepMs)
+
+		desc := fmt.Sprintf("case=%d step=%d lookback=%d start=%d end=%d samples=%d",
+			caseIdx, stepMs, lookbackMs, queryStart, queryEnd, len(samples))
+
+		for evalTs := queryStart; evalTs <= queryEnd; evalTs += stepMs {
+			rawPick, rawOK := engineSelect(samples, evalTs, lookbackMs)
+			reducedPick, reducedOK := engineSelect(reduced, evalTs, lookbackMs)
+
+			require.Equal(t, rawOK, reducedOK, "%s eval=%d presence mismatch", desc, evalTs)
+			if rawOK {
+				require.Equal(t, rawPick, reducedPick, "%s eval=%d sample mismatch", desc, evalTs)
+			}
+		}
+	}
+}
+
+// Instant queries (step 0) evaluate once at firstEval == hints.End; lastSamplePerStep
+// collapses to a single bucket over the whole window.
+func TestLastSamplePerStepEquivalenceInstantQuery(t *testing.T) {
+	rng := rand.New(rand.NewSource(7))
+
+	for caseIdx := 0; caseIdx < 500; caseIdx++ {
+		lookbackMs := []int64{1, 3, 5, 10, 45}[rng.Intn(5)]
+		evalTs := int64(1000)
+		hintsStart := evalTs - (lookbackMs - 1)
+		hintsEnd := evalTs
+		firstEval := hintsStart + lookbackMs - 1
+		require.Equal(t, evalTs, firstEval)
+
+		windowSize := hintsEnd - hintsStart + 1
+		numSamples := rng.Int63n(10)
+		if numSamples > windowSize {
+			numSamples = windowSize
+		}
+		seen := make(map[int64]bool)
+		var samples []tsample
+		for int64(len(samples)) < numSamples {
+			ts := hintsStart + rng.Int63n(windowSize)
+			if seen[ts] {
+				continue
+			}
+			seen[ts] = true
+			samples = append(samples, tsample{ts: ts, value: rng.Float64(), stale: rng.Intn(4) == 0})
+		}
+		sort.Slice(samples, func(i, j int) bool { return samples[i].ts < samples[j].ts })
+
+		reduced := lastPerStep(samples, firstEval, 0)
+		require.LessOrEqual(t, len(reduced), 1, "instant reduction must keep at most one sample")
+
+		rawPick, rawOK := engineSelect(samples, evalTs, lookbackMs)
+		reducedPick, reducedOK := engineSelect(reduced, evalTs, lookbackMs)
+		require.Equal(t, rawOK, reducedOK, "case=%d presence mismatch", caseIdx)
+		if rawOK {
+			require.Equal(t, rawPick, reducedPick, "case=%d sample mismatch", caseIdx)
+		}
+	}
+}
+
+// A stale marker that is the latest sample of its bucket must shadow older
+// samples: the engine sees the marker and reports the series absent, exactly
+// as with raw samples. Pre-filtering stale rows would instead resurrect the
+// older sample.
+func TestLastSamplePerStepKeepsStaleShadowing(t *testing.T) {
+	lookbackMs := int64(10)
+	stepMs := int64(5)
+	queryStart := int64(1000)
+
+	samples := []tsample{
+		{ts: 998, value: 1.0},                // bucket 0
+		{ts: 999, stale: true},               // bucket 0: marker shadows 998
+		{ts: 1003, value: 2.0},               // bucket 1
+		{ts: 1004, stale: true},              // bucket 1: marker shadows 1003
+		{ts: 1008, value: 3.0, stale: false}, // bucket 2
+	}
+	firstEval := queryStart
+	reduced := lastPerStep(samples, firstEval, stepMs)
+
+	for evalTs := queryStart; evalTs <= queryStart+2*stepMs; evalTs += stepMs {
+		rawPick, rawOK := engineSelect(samples, evalTs, lookbackMs)
+		reducedPick, reducedOK := engineSelect(reduced, evalTs, lookbackMs)
+		require.Equal(t, rawOK, reducedOK, "eval=%d", evalTs)
+		if rawOK {
+			require.Equal(t, rawPick, reducedPick, "eval=%d", evalTs)
+		}
+	}
+}

pkg/prometheus/clickhouseprometheusv2/provider.go

@@ -0,0 +1,84 @@
+package clickhouseprometheusv2
+
+import (
+	"context"
+	"time"
+
+	"github.com/SigNoz/signoz/pkg/factory"
+	"github.com/SigNoz/signoz/pkg/prometheus"
+	"github.com/SigNoz/signoz/pkg/telemetrystore"
+	"github.com/prometheus/prometheus/promql"
+	"github.com/prometheus/prometheus/storage"
+)
+
+// Provider ties the package together: its own engine and parser, the
+// ClickHouse client behind the native storage.Querier, and the transpiler
+// executor. See the package documentation for what runs where and why. It is
+// exported as a concrete type — pkg/querier holds it directly for shadow
+// comparison and pinned serving, and an interface with a single
+// implementation would only hide that dependency.
+type Provider struct {
+	settings factory.ScopedProviderSettings
+	engine   *prometheus.Engine
+	parser   prometheus.Parser
+	client   *client
+	executor *executor
+}
+
+var (
+	_ prometheus.Prometheus        = (*Provider)(nil)
+	_ prometheus.StatementCapturer = (*Provider)(nil)
+)
+
+func NewFactory(telemetryStore telemetrystore.TelemetryStore) factory.ProviderFactory[prometheus.Prometheus, prometheus.Config] {
+	return factory.NewProviderFactory(factory.MustNewName("clickhousev2"), func(ctx context.Context, providerSettings factory.ProviderSettings, config prometheus.Config) (prometheus.Prometheus, error) {
+		return New(ctx, providerSettings, config, telemetryStore)
+	})
+}
+
+func New(_ context.Context, providerSettings factory.ProviderSettings, config prometheus.Config, telemetryStore telemetrystore.TelemetryStore) (*Provider, error) {
+	settings := factory.NewScopedProviderSettings(providerSettings, "github.com/SigNoz/signoz/pkg/prometheus/clickhouseprometheusv2")
+
+	engine := prometheus.NewEngine(settings.Logger(), config)
+	parser := prometheus.NewParser()
+	client := newClient(settings, telemetryStore, config)
+
+	return &Provider{
+		settings: settings,
+		engine:   engine,
+		parser:   parser,
+		client:   client,
+		executor: &executor{client: client, engine: engine, parser: parser},
+	}, nil
+}
+
+// TryExecuteRange evaluates transpilable query shapes directly in ClickHouse
+// (see transpiler.go). ok=false means the shape is not transpilable and the
+// caller should evaluate through Engine over Storage instead.
+func (p *Provider) TryExecuteRange(ctx context.Context, query string, start, end time.Time, step time.Duration) (promql.Matrix, bool, error) {
+	return p.executor.TryExecuteRange(ctx, query, start, end, step)
+}
+
+func (p *Provider) Engine() *prometheus.Engine {
+	return p.engine
+}
+
+func (p *Provider) Parser() prometheus.Parser {
+	return p.parser
+}
+
+func (p *Provider) Storage() storage.Queryable {
+	return p
+}
+
+func (p *Provider) Querier(mint, maxt int64) (storage.Querier, error) {
+	return &querier{mint: mint, maxt: maxt, client: p.client}, nil
+}
+
+// CapturingStorage implements prometheus.StatementCapturer: a storage that
+// records each selector's SQL without executing it, for the preview path.
+// A fresh recorder per call keeps concurrent dry-runs isolated.
+func (p *Provider) CapturingStorage() (storage.Queryable, prometheus.StatementRecorder) {
+	recorder := &statementRecorder{}
+	return &captureQueryable{client: p.client, recorder: recorder}, recorder
+}

pkg/prometheus/clickhouseprometheusv2/querier.go

@@ -0,0 +1,233 @@
+package clickhouseprometheusv2
+
+import (
+	"context"
+	"fmt"
+	"slices"
+	"sort"
+	"time"
+
+	"github.com/SigNoz/signoz/pkg/prometheus"
+	"github.com/huandu/go-sqlbuilder"
+	"github.com/prometheus/prometheus/model/labels"
+	"github.com/prometheus/prometheus/storage"
+	"github.com/prometheus/prometheus/util/annotations"
+)
+
+// defaultLookbackDelta mirrors promql's default when the config leaves the
+// lookback unset; the engine and the storage must agree on it for
+// last-sample-per-step bucket anchoring.
+const defaultLookbackDelta = 5 * time.Minute
+
+// querier is a native storage.Querier over ClickHouse. Unlike v1 it does not
+// round-trip through the remote-read protobuf machinery: Select builds SQL
+// directly from the matchers and hints, and the result set is assembled once
+// into compact series.
+type querier struct {
+	mint, maxt int64
+	client     *client
+}
+
+var _ storage.Querier = (*querier)(nil)
+
+func (q *querier) Select(ctx context.Context, sortSeries bool, hints *storage.SelectHints, matchers ...*labels.Matcher) storage.SeriesSet {
+	if rawQuery, ok := rawSQLQuery(matchers); ok {
+		_, end := q.window(hints)
+		list, err := q.client.queryRaw(ctx, rawQuery, end)
+		if err != nil {
+			return storage.ErrSeriesSet(err)
+		}
+		if sortSeries {
+			sort.Slice(list, func(i, j int) bool { return labels.Compare(list[i].lset, list[j].lset) < 0 })
+		}
+		return newSeriesSet(list)
+	}
+
+	start, end := q.window(hints)
+
+	seriesQuery, seriesArgs, err := buildSeriesQuery(start, end, matchers)
+	if err != nil {
+		return storage.ErrSeriesSet(err)
+	}
+	lookup, err := q.client.selectSeries(ctx, seriesQuery, seriesArgs)
+	if err != nil {
+		return storage.ErrSeriesSet(err)
+	}
+	if len(lookup.fingerprints) == 0 {
+		return storage.EmptySeriesSet()
+	}
+
+	list, err := q.fetchSamples(ctx, start, end, matchers, lookup, q.lastSamplePerStepFor(ctx, hints))
+	if err != nil {
+		return storage.ErrSeriesSet(err)
+	}
+
+	// Sorting doubles as duplicate-label-set detection, which the engine
+	// depends on storages never emitting; the cost is on series count, not
+	// samples.
+	list = sortAndMerge(list)
+	return newSeriesSet(list)
+}
+
+// LabelValues returns the values of a label across series matching the
+// matchers within the querier window.
+func (q *querier) LabelValues(ctx context.Context, name string, hints *storage.LabelHints, matchers ...*labels.Matcher) ([]string, annotations.Annotations, error) {
+	sb := sqlbuilder.NewSelectBuilder()
+	if name == metricNameLabel {
+		sb.Select("DISTINCT metric_name AS value")
+	} else {
+		sb.Select(fmt.Sprintf("DISTINCT JSONExtractString(labels, %s) AS value", sb.Var(name)))
+	}
+	adjustedStart, table := timeSeriesTableFor(q.mint, q.maxt)
+	sb.From(fmt.Sprintf("%s.%s", databaseName, table))
+	if err := applySeriesConditions(sb, adjustedStart, q.maxt, matchers); err != nil {
+		return nil, nil, err
+	}
+	sb.Where("value != ''")
+	if hints != nil && hints.Limit > 0 {
+		sb.Limit(hints.Limit)
+	}
+
+	query, args := sb.BuildWithFlavor(sqlbuilder.ClickHouse)
+	values, err := q.selectStrings(ctx, "LabelValues", query, args)
+	if err != nil {
+		return nil, nil, err
+	}
+	slices.Sort(values)
+	return values, nil, nil
+}
+
+// LabelNames returns the label names present on series matching the matchers
+// within the querier window.
+func (q *querier) LabelNames(ctx context.Context, hints *storage.LabelHints, matchers ...*labels.Matcher) ([]string, annotations.Annotations, error) {
+	sb := sqlbuilder.NewSelectBuilder()
+	sb.Select("DISTINCT arrayJoin(JSONExtractKeys(labels)) AS name")
+	adjustedStart, table := timeSeriesTableFor(q.mint, q.maxt)
+	sb.From(fmt.Sprintf("%s.%s", databaseName, table))
+	if err := applySeriesConditions(sb, adjustedStart, q.maxt, matchers); err != nil {
+		return nil, nil, err
+	}
+	if hints != nil && hints.Limit > 0 {
+		sb.Limit(hints.Limit)
+	}
+
+	query, args := sb.BuildWithFlavor(sqlbuilder.ClickHouse)
+	names, err := q.selectStrings(ctx, "LabelNames", query, args)
+	if err != nil {
+		return nil, nil, err
+	}
+	slices.Sort(names)
+	return names, nil, nil
+}
+
+func (q *querier) Close() error {
+	return nil
+}
+
+// window returns the per-selector fetch window. The engine sends per-selector
+// bounds in the hints (already adjusted for offset, @, range and lookback);
+// they are always at least as tight as the querier-level mint/maxt, which
+// span the union of all selectors in the query.
+func (q *querier) window(hints *storage.SelectHints) (int64, int64) {
+	if hints != nil && hints.Start != 0 && hints.End != 0 && hints.Start <= hints.End {
+		return hints.Start, hints.End
+	}
+	return q.mint, q.maxt
+}
+
+// lastSamplePerStepFor decides whether the fetch can keep only the last
+// sample per step bucket, and computes the bucket parameters. Requirements:
+//   - the call site attached QueryTraits proving the query has no subquery
+//     (subquery selectors evaluate at the subquery's own step, but hints
+//     carry the top-level step);
+//   - the selector is an instant selector (hints.Range == 0); range selectors
+//     need every raw sample in the window;
+//   - per-selector hints are present.
+//
+// The engine derives hints.Start for instant selectors as
+// firstEval - (lookback - 1ms), so the first evaluation timestamp is
+// recovered as hints.Start + lookback - 1ms. Bucket boundaries then coincide
+// with evaluation timestamps, which is what makes keeping only the last
+// sample per bucket lossless.
+func (q *querier) lastSamplePerStepFor(ctx context.Context, hints *storage.SelectHints) *lastSamplePerStep {
+	if hints == nil || hints.Range != 0 || hints.Start <= 0 {
+		return nil
+	}
+	traits, ok := prometheus.QueryTraitsFromContext(ctx)
+	if !ok || !traits.SubqueryFree {
+		return nil
+	}
+	firstEval := hints.Start + q.client.lookbackMs - 1
+	if firstEval > hints.End {
+		// Defensive: never anchor a bucket past the window.
+		firstEval = hints.End
+	}
+	return &lastSamplePerStep{firstEvalMs: firstEval, stepMs: hints.Step}
+}
+
+// fetchSamples runs the samples query for the matched series. Small sets
+// inline the fingerprints as sorted uint64 literals — literals engage the
+// samples primary key, and sorting keeps the statement deterministic for
+// logging and tests. Larger sets re-run the series predicates as a
+// shard-local IN subquery instead: inlining hundreds of thousands of
+// literals makes the statement itself the bottleneck, while the subquery is
+// a cheap primary-key scan on each shard's own series table (see
+// localTimeSeriesTable for why that is complete).
+func (q *querier) fetchSamples(ctx context.Context, start, end int64, matchers []*labels.Matcher, lookup *seriesLookup, lastPerStep *lastSamplePerStep) ([]*series, error) {
+	var fingerprints []uint64
+	if len(lookup.fingerprints) <= inlineFingerprintsLimit {
+		fingerprints = make([]uint64, 0, len(lookup.fingerprints))
+		for fp := range lookup.fingerprints {
+			fingerprints = append(fingerprints, fp)
+		}
+		slices.Sort(fingerprints)
+	}
+	query, args, err := buildSamplesQuery(start, end, lookup.metricNames, fingerprints, matchers, lastPerStep)
+	if err != nil {
+		return nil, err
+	}
+	return q.client.selectSamples(ctx, query, args, lookup)
+}
+
+func (q *querier) selectStrings(ctx context.Context, fn, query string, args []any) ([]string, error) {
+	ctx = q.client.withContext(ctx, fn)
+	rows, err := q.client.telemetryStore.ClickhouseDB().Query(ctx, query, args...)
+	if err != nil {
+		return nil, err
+	}
+	defer rows.Close()
+
+	var out []string
+	var v string
+	for rows.Next() {
+		if err := rows.Scan(&v); err != nil {
+			return nil, err
+		}
+		out = append(out, v)
+	}
+	if err := rows.Err(); err != nil {
+		return nil, err
+	}
+	return out, nil
+}
+
+// rawSQLQuery detects the {job="rawsql", query="..."} escape hatch.
+func rawSQLQuery(matchers []*labels.Matcher) (string, bool) {
+	if len(matchers) != 2 {
+		return "", false
+	}
+	var hasJob bool
+	var query string
+	for _, m := range matchers {
+		if m.Type == labels.MatchEqual && m.Name == "job" && m.Value == "rawsql" {
+			hasJob = true
+		}
+		if m.Type == labels.MatchEqual && m.Name == "query" {
+			query = m.Value
+		}
+	}
+	if hasJob && query != "" {
+		return query, true
+	}
+	return "", false
+}

pkg/prometheus/clickhouseprometheusv2/querier_test.go

@@ -0,0 +1,221 @@
+package clickhouseprometheusv2
+
+import (
+	"context"
+	"fmt"
+	"testing"
+
+	"github.com/DATA-DOG/go-sqlmock"
+	cmock "github.com/SigNoz/clickhouse-go-mock"
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+
+	"github.com/SigNoz/signoz/pkg/errors"
+	"github.com/SigNoz/signoz/pkg/factory"
+	"github.com/SigNoz/signoz/pkg/instrumentation/instrumentationtest"
+	"github.com/SigNoz/signoz/pkg/prometheus"
+	"github.com/SigNoz/signoz/pkg/telemetrystore"
+	"github.com/SigNoz/signoz/pkg/telemetrystore/telemetrystoretest"
+	"github.com/prometheus/prometheus/model/labels"
+	"github.com/prometheus/prometheus/storage"
+)
+
+var (
+	seriesCols = []cmock.ColumnType{
+		{Name: "fingerprint", Type: "UInt64"},
+		{Name: "labels", Type: "String"},
+	}
+	samplesCols = []cmock.ColumnType{
+		{Name: "fingerprint", Type: "UInt64"},
+		{Name: "unix_milli", Type: "Int64"},
+		{Name: "value", Type: "Float64"},
+		{Name: "flags", Type: "UInt32"},
+	}
+)
+
+func newTestClient(t *testing.T, cfg prometheus.ClickhouseV2Config) (*client, *telemetrystoretest.Provider) {
+	t.Helper()
+	store := telemetrystoretest.New(telemetrystore.Config{Provider: "clickhouse"}, sqlmock.QueryMatcherRegexp)
+	settings := factory.NewScopedProviderSettings(instrumentationtest.New().ToProviderSettings(), "clickhouseprometheusv2_test")
+	promCfg := prometheus.Config{ClickhouseV2: cfg}
+	return newClient(settings, store, promCfg), store
+}
+
+func testMatchers(t *testing.T) []*labels.Matcher {
+	t.Helper()
+	return []*labels.Matcher{
+		mustMatcher(t, labels.MatchEqual, "__name__", "cpu_usage"),
+		mustMatcher(t, labels.MatchEqual, "job", "api"),
+	}
+}
+
+func TestQuerierSelectRawPath(t *testing.T) {
+	c, store := newTestClient(t, prometheus.ClickhouseV2Config{})
+	q := &querier{mint: 1000, maxt: 2000, client: c}
+
+	store.Mock().ExpectQuery("SELECT fingerprint, any\\(labels\\)").WithArgs("cpu_usage", int64(0), int64(2000), "job", "api").WillReturnRows(cmock.NewRows(seriesCols, [][]any{
+		{uint64(42), `{"__name__":"cpu_usage","job":"api","instance":"a"}`},
+		{uint64(7), `{"__name__":"cpu_usage","job":"api","instance":"b"}`},
+	}))
+	// Inline fingerprints (sorted), raw samples: no traits in ctx -> no
+	// last-sample-per-step reduction.
+	store.Mock().ExpectQuery("SELECT fingerprint, unix_milli, value, flags FROM signoz_metrics.distributed_samples_v4 WHERE metric_name = \\? AND temporality IN \\['Cumulative', 'Unspecified'\\] AND fingerprint IN \\(7, 42\\)").
+		WithArgs("cpu_usage", int64(1000), int64(2000)).
+		WillReturnRows(cmock.NewRows(samplesCols, [][]any{
+			{uint64(7), int64(1100), 1.5, uint32(0)},
+			{uint64(7), int64(1200), 2.5, uint32(0)},
+			{uint64(42), int64(1100), 3.5, uint32(1)}, // stale marker
+		}))
+
+	hints := &storage.SelectHints{Start: 1000, End: 2000, Step: 60_000}
+	set := q.Select(context.Background(), false, hints, testMatchers(t)...)
+
+	var got []*series
+	for set.Next() {
+		got = append(got, set.At().(*series))
+	}
+	require.NoError(t, set.Err())
+	require.Len(t, got, 2)
+
+	// Sorted by labels: instance=a (fp 42) before instance=b (fp 7).
+	assert.Equal(t, "a", got[0].lset.Get("instance"))
+	require.Len(t, got[0].ts, 1)
+	assert.True(t, got[0].vs[0] != got[0].vs[0], "stale marker must be NaN") //nolint:testifylint
+
+	assert.Equal(t, "b", got[1].lset.Get("instance"))
+	assert.Equal(t, []int64{1100, 1200}, got[1].ts)
+	assert.Equal(t, []float64{1.5, 2.5}, got[1].vs)
+
+	// No fingerprint label injected.
+	assert.Empty(t, got[0].lset.Get("fingerprint"))
+}
+
+// Wrong gating silently corrupts range functions (a rate over reduced
+// samples loses points), so the decision logic is pinned here even though
+// the helper is unexported: the integration suite would catch it too, but
+// with far worse failure locality.
+func TestLastSamplePerStepFor(t *testing.T) {
+	c, _ := newTestClient(t, prometheus.ClickhouseV2Config{})
+	q := &querier{mint: 0, maxt: 2000, client: c}
+	traitsCtx := prometheus.NewContextWithQueryTraits(context.Background(), prometheus.QueryTraits{SubqueryFree: true})
+
+	tests := []struct {
+		name  string
+		ctx   context.Context
+		hints *storage.SelectHints
+		want  *lastSamplePerStep
+	}{
+		{"no traits in context stays raw", context.Background(), &storage.SelectHints{Start: 1000, End: 2000, Step: 60_000}, nil},
+		{"subquery in the query stays raw", prometheus.NewContextWithQueryTraits(context.Background(), prometheus.QueryTraits{SubqueryFree: false}), &storage.SelectHints{Start: 1000, End: 2000, Step: 60_000}, nil},
+		{"range selector stays raw", traitsCtx, &storage.SelectHints{Start: 1000, End: 2000, Step: 60_000, Range: 300_000}, nil},
+		{"instant selector reduces, anchored at first eval", traitsCtx, &storage.SelectHints{Start: 1000, End: 2_000_000, Step: 60_000}, &lastSamplePerStep{firstEvalMs: 1000 + c.lookbackMs - 1, stepMs: 60_000}},
+		{"anchor never passes the window end", traitsCtx, &storage.SelectHints{Start: 1000, End: 2000, Step: 60_000}, &lastSamplePerStep{firstEvalMs: 2000, stepMs: 60_000}},
+	}
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			assert.Equal(t, tt.want, q.lastSamplePerStepFor(tt.ctx, tt.hints))
+		})
+	}
+}
+
+func TestQuerierSelectSeriesBudget(t *testing.T) {
+	c, store := newTestClient(t, prometheus.ClickhouseV2Config{MaxFetchedSeries: 1})
+	q := &querier{mint: 1000, maxt: 2000, client: c}
+
+	store.Mock().ExpectQuery("SELECT fingerprint, any\\(labels\\)").WithArgs("cpu_usage", int64(0), int64(2000), "job", "api").WillReturnRows(cmock.NewRows(seriesCols, [][]any{
+		{uint64(1), `{"__name__":"cpu_usage","instance":"a"}`},
+		{uint64(2), `{"__name__":"cpu_usage","instance":"b"}`},
+	}))
+
+	set := q.Select(context.Background(), false, &storage.SelectHints{Start: 1000, End: 2000}, testMatchers(t)...)
+	assert.False(t, set.Next())
+	require.Error(t, set.Err())
+	assert.True(t, errors.Ast(set.Err(), errors.TypeInvalidInput), "budget error must be typed invalid input, got %v", set.Err())
+}
+
+func TestQuerierSelectSamplesBudget(t *testing.T) {
+	c, store := newTestClient(t, prometheus.ClickhouseV2Config{MaxFetchedSamples: 2})
+	q := &querier{mint: 1000, maxt: 2000, client: c}
+
+	store.Mock().ExpectQuery("SELECT fingerprint, any\\(labels\\)").WithArgs("cpu_usage", int64(0), int64(2000)).WillReturnRows(cmock.NewRows(seriesCols, [][]any{
+		{uint64(7), `{"__name__":"cpu_usage"}`},
+	}))
+	store.Mock().ExpectQuery("SELECT fingerprint, unix_milli, value, flags").
+		WithArgs("cpu_usage", int64(1000), int64(2000)).
+		WillReturnRows(cmock.NewRows(samplesCols, [][]any{
+			{uint64(7), int64(1100), 1.0, uint32(0)},
+			{uint64(7), int64(1200), 2.0, uint32(0)},
+			{uint64(7), int64(1300), 3.0, uint32(0)},
+		}))
+
+	set := q.Select(context.Background(), false, &storage.SelectHints{Start: 1000, End: 2000},
+		mustMatcher(t, labels.MatchEqual, "__name__", "cpu_usage"))
+	assert.False(t, set.Next())
+	require.Error(t, set.Err())
+	assert.True(t, errors.Ast(set.Err(), errors.TypeInvalidInput))
+}
+
+func TestQuerierSelectSubqueryFilterOverInlineLimit(t *testing.T) {
+	c, store := newTestClient(t, prometheus.ClickhouseV2Config{})
+	q := &querier{mint: 1000, maxt: 2000, client: c}
+
+	seriesRows := make([][]any, inlineFingerprintsLimit+1)
+	for i := range seriesRows {
+		seriesRows[i] = []any{uint64(i + 1), fmt.Sprintf(`{"__name__":"cpu_usage","instance":"i%d"}`, i)}
+	}
+	store.Mock().ExpectQuery("SELECT fingerprint, any\\(labels\\)").WithArgs("cpu_usage", int64(0), int64(2000), "job", "api").WillReturnRows(cmock.NewRows(seriesCols, seriesRows))
+	// The over-limit samples query embeds the semi-join against the
+	// shard-local series table (fingerprint co-locality), not a GLOBAL
+	// broadcast; args follow placeholder order — samples metric name, then
+	// the semi-join's series predicates, then the samples window bounds.
+	store.Mock().ExpectQuery("fingerprint IN \\(SELECT fingerprint FROM signoz_metrics\\.time_series_v4").
+		WithArgs("cpu_usage", "cpu_usage", int64(0), int64(2000), "job", "api", int64(1000), int64(2000)).
+		WillReturnRows(cmock.NewRows(samplesCols, [][]any{}))
+
+	set := q.Select(context.Background(), false, &storage.SelectHints{Start: 1000, End: 2000}, testMatchers(t)...)
+	assert.False(t, set.Next())
+	require.NoError(t, set.Err())
+}
+
+func TestQuerierSelectRawSQLPassthrough(t *testing.T) {
+	c, store := newTestClient(t, prometheus.ClickhouseV2Config{})
+	q := &querier{mint: 1000, maxt: 2000, client: c}
+
+	rawCols := []cmock.ColumnType{
+		{Name: "le", Type: "String"},
+		{Name: "value", Type: "Float64"},
+	}
+	store.Mock().ExpectQuery("SELECT le, avg\\(v\\) AS value FROM t").WillReturnRows(cmock.NewRows(rawCols, [][]any{
+		{"0.5", 12.5},
+	}))
+
+	set := q.Select(context.Background(), false, &storage.SelectHints{Start: 1000, End: 2000},
+		mustMatcher(t, labels.MatchEqual, "job", "rawsql"),
+		mustMatcher(t, labels.MatchEqual, "query", "SELECT le, avg(v) AS value FROM t"),
+	)
+
+	require.True(t, set.Next())
+	s := set.At()
+	assert.Equal(t, "0.5", s.Labels().Get("le"))
+	it := s.Iterator(nil)
+	require.NotNil(t, it)
+	_, v := func() (int64, float64) { it.Next(); return it.At() }()
+	assert.Equal(t, 12.5, v)
+	assert.False(t, set.Next())
+}
+
+func TestCaptureQuerierRecordsWithoutExecuting(t *testing.T) {
+	c, _ := newTestClient(t, prometheus.ClickhouseV2Config{})
+	recorder := &statementRecorder{}
+	cq := &captureQuerier{querier: querier{mint: 1000, maxt: 2000, client: c}, recorder: recorder}
+
+	ctx := prometheus.NewContextWithQueryTraits(context.Background(), prometheus.QueryTraits{SubqueryFree: true})
+	set := cq.Select(ctx, false, &storage.SelectHints{Start: 1000, End: 2000, Step: 60_000}, testMatchers(t)...)
+	assert.False(t, set.Next())
+	require.NoError(t, set.Err())
+
+	statements := recorder.Statements()
+	require.Len(t, statements, 1)
+	assert.Contains(t, statements[0].Query, "IN (SELECT fingerprint FROM signoz_metrics.time_series_v4")
+	assert.Contains(t, statements[0].Query, "argMax(value, unix_milli)")
+}

pkg/prometheus/clickhouseprometheusv2/seriesset.go

@@ -0,0 +1,184 @@
+package clickhouseprometheusv2
+
+import (
+	"sort"
+
+	"github.com/prometheus/prometheus/model/histogram"
+	"github.com/prometheus/prometheus/model/labels"
+	"github.com/prometheus/prometheus/storage"
+	"github.com/prometheus/prometheus/tsdb/chunkenc"
+	"github.com/prometheus/prometheus/util/annotations"
+)
+
+// series is one time series with samples stored as parallel slices, ordered
+// by timestamp. The compact layout avoids per-sample allocations and keeps
+// iteration cache friendly.
+type series struct {
+	lset labels.Labels
+	ts   []int64
+	vs   []float64
+}
+
+var _ storage.Series = (*series)(nil)
+
+func (s *series) Labels() labels.Labels {
+	return s.lset
+}
+
+func (s *series) Iterator(it chunkenc.Iterator) chunkenc.Iterator {
+	if fit, ok := it.(*floatIterator); ok {
+		fit.reset(s)
+		return fit
+	}
+	fit := &floatIterator{}
+	fit.reset(s)
+	return fit
+}
+
+// floatIterator implements chunkenc.Iterator over a series' sample slices.
+type floatIterator struct {
+	s *series
+	i int
+}
+
+var _ chunkenc.Iterator = (*floatIterator)(nil)
+
+func (it *floatIterator) reset(s *series) {
+	it.s = s
+	it.i = -1
+}
+
+func (it *floatIterator) Next() chunkenc.ValueType {
+	it.i++
+	if it.i >= len(it.s.ts) {
+		return chunkenc.ValNone
+	}
+	return chunkenc.ValFloat
+}
+
+func (it *floatIterator) Seek(t int64) chunkenc.ValueType { //nolint:govet // stdmethods flags io.Seeker; this is chunkenc.Iterator's Seek
+	if it.i < 0 {
+		it.i = 0
+	}
+	if it.i >= len(it.s.ts) {
+		return chunkenc.ValNone
+	}
+	// The current position, once valid, must not move backwards.
+	if it.s.ts[it.i] >= t {
+		return chunkenc.ValFloat
+	}
+	it.i += sort.Search(len(it.s.ts)-it.i, func(j int) bool {
+		return it.s.ts[it.i+j] >= t
+	})
+	if it.i >= len(it.s.ts) {
+		return chunkenc.ValNone
+	}
+	return chunkenc.ValFloat
+}
+
+func (it *floatIterator) At() (int64, float64) {
+	return it.s.ts[it.i], it.s.vs[it.i]
+}
+
+func (it *floatIterator) AtHistogram(*histogram.Histogram) (int64, *histogram.Histogram) {
+	return 0, nil
+}
+
+func (it *floatIterator) AtFloatHistogram(*histogram.FloatHistogram) (int64, *histogram.FloatHistogram) {
+	return 0, nil
+}
+
+func (it *floatIterator) AtT() int64 {
+	return it.s.ts[it.i]
+}
+
+// AtST returns the current start timestamp; not tracked by this storage.
+func (it *floatIterator) AtST() int64 {
+	return 0
+}
+
+func (it *floatIterator) Err() error {
+	return nil
+}
+
+// seriesSet iterates a fully materialized, label-sorted list of series.
+type seriesSet struct {
+	series []*series
+	i      int
+}
+
+var _ storage.SeriesSet = (*seriesSet)(nil)
+
+func newSeriesSet(list []*series) *seriesSet {
+	return &seriesSet{series: list, i: -1}
+}
+
+func (s *seriesSet) Next() bool {
+	s.i++
+	return s.i < len(s.series)
+}
+
+func (s *seriesSet) At() storage.Series {
+	return s.series[s.i]
+}
+
+func (s *seriesSet) Err() error {
+	return nil
+}
+
+func (s *seriesSet) Warnings() annotations.Annotations {
+	return nil
+}
+
+// sortAndMerge orders series by label set and merges series whose label sets
+// are identical. Distinct fingerprints can carry identical label sets (e.g.
+// series differing only in a non-label dimension); Prometheus storages never
+// expose duplicate label sets to the engine, so merge their samples by
+// timestamp, keeping the first sample on ties.
+func sortAndMerge(list []*series) []*series {
+	if len(list) < 2 {
+		return list
+	}
+	sort.Slice(list, func(i, j int) bool {
+		return labels.Compare(list[i].lset, list[j].lset) < 0
+	})
+	out := list[:1]
+	for _, s := range list[1:] {
+		last := out[len(out)-1]
+		if labels.Compare(last.lset, s.lset) != 0 {
+			out = append(out, s)
+			continue
+		}
+		merged := mergeSamples(last, s)
+		out[len(out)-1] = merged
+	}
+	return out
+}
+
+func mergeSamples(a, b *series) *series {
+	ts := make([]int64, 0, len(a.ts)+len(b.ts))
+	vs := make([]float64, 0, len(a.ts)+len(b.ts))
+	i, j := 0, 0
+	for i < len(a.ts) && j < len(b.ts) {
+		switch {
+		case a.ts[i] < b.ts[j]:
+			ts = append(ts, a.ts[i])
+			vs = append(vs, a.vs[i])
+			i++
+		case a.ts[i] > b.ts[j]:
+			ts = append(ts, b.ts[j])
+			vs = append(vs, b.vs[j])
+			j++
+		default:
+			ts = append(ts, a.ts[i])
+			vs = append(vs, a.vs[i])
+			i++
+			j++
+		}
+	}
+	ts = append(ts, a.ts[i:]...)
+	vs = append(vs, a.vs[i:]...)
+	ts = append(ts, b.ts[j:]...)
+	vs = append(vs, b.vs[j:]...)
+	return &series{lset: a.lset, ts: ts, vs: vs}
+}

pkg/prometheus/clickhouseprometheusv2/sql.go

@@ -0,0 +1,199 @@
+package clickhouseprometheusv2
+
+import (
+	"fmt"
+	"strconv"
+	"strings"
+
+	"github.com/SigNoz/signoz/pkg/errors"
+	"github.com/SigNoz/signoz/pkg/query-service/constants"
+	"github.com/huandu/go-sqlbuilder"
+	"github.com/prometheus/prometheus/model/labels"
+)
+
+// inlineFingerprintsLimit is the largest matched-series count inlined into
+// the samples query as literals. Literals engage the samples primary key and
+// avoid a second series-table scan; past a few thousand the statement itself
+// becomes the cost, and the shard-local subquery filter wins. Not
+// configurable: the crossover depends on statement parsing, not on any
+// property of a deployment an operator could know better.
+const inlineFingerprintsLimit = 5_000
+
+// buildSeriesQuery renders the series lookup: one row per matched fingerprint
+// with its labels.
+func buildSeriesQuery(start, end int64, matchers []*labels.Matcher) (string, []any, error) {
+	adjustedStart, table := timeSeriesTableFor(start, end)
+	sb := sqlbuilder.NewSelectBuilder()
+	sb.Select("fingerprint", "any(labels)")
+	sb.From(fmt.Sprintf("%s.%s", databaseName, table))
+	if err := applySeriesConditions(sb, adjustedStart, end, matchers); err != nil {
+		return "", nil, err
+	}
+	sb.GroupBy("fingerprint")
+	query, args := sb.BuildWithFlavor(sqlbuilder.ClickHouse)
+	return query, args, nil
+}
+
+// buildSamplesQuery renders the samples fetch for the series selected by the
+// series lookup. Small matched sets pass inlineFingerprints — sorted uint64
+// literals that engage the samples primary key; nil means the set exceeded
+// the inline limit, and the filter becomes a semi-join re-running the series
+// predicates against the shard-local series table (complete by fingerprint
+// co-locality, see localTimeSeriesTable; a GLOBAL broadcast of the matched
+// set would ship it to every shard instead). metricNames narrows the
+// primary-key scan; when the selector had no __name__ equality, the names
+// observed on the matched series are used. A non-nil lastPerStep groups to
+// one (the last) sample per step bucket.
+func buildSamplesQuery(start, end int64, metricNames []string, inlineFingerprints []uint64, matchers []*labels.Matcher, lastPerStep *lastSamplePerStep) (string, []any, error) {
+	sb := sqlbuilder.NewSelectBuilder()
+	if lastPerStep != nil {
+		// Aliases must not shadow source columns: ClickHouse resolves aliases
+		// in WHERE too, and "max(unix_milli) AS unix_milli" would put an
+		// aggregate into the WHERE clause (error 184).
+		sb.Select("fingerprint", "max(unix_milli) AS ts", "argMax(value, unix_milli) AS val", "argMax(flags, unix_milli) AS fl")
+	} else {
+		sb.Select("fingerprint", "unix_milli", "value", "flags")
+	}
+	sb.From(fmt.Sprintf("%s.%s", databaseName, distributedSamplesV4))
+
+	switch len(metricNames) {
+	case 0:
+		// No name constraint derivable; correct but unable to use the
+		// metric_name primary-key prefix.
+	case 1:
+		sb.Where(sb.EQ("metric_name", metricNames[0]))
+	default:
+		sb.Where(sb.In("metric_name", sqlbuilder.List(metricNames)))
+	}
+	// temporality precedes metric_name in the samples primary key; the
+	// fingerprints already come from these temporalities, so this only helps
+	// granule pruning.
+	sb.Where("temporality IN ['Cumulative', 'Unspecified']")
+	if inlineFingerprints != nil {
+		sb.Where("fingerprint " + inlineFingerprintFilter(inlineFingerprints))
+	} else {
+		sub := sqlbuilder.NewSelectBuilder()
+		sub.Select("fingerprint")
+		adjustedStart, table := timeSeriesTableFor(start, end)
+		sub.From(fmt.Sprintf("%s.%s", databaseName, localTimeSeriesTable(table)))
+		if err := applySeriesConditions(sub, adjustedStart, end, matchers); err != nil {
+			return "", nil, err
+		}
+		sb.Where(sb.In("fingerprint", sub))
+	}
+	sb.Where(sb.GTE("unix_milli", start), sb.LTE("unix_milli", end))
+
+	if lastPerStep != nil {
+		sb.GroupBy("fingerprint")
+		if expr := lastPerStep.bucketExpr(); expr != "" {
+			sb.GroupBy(expr)
+		}
+		sb.OrderBy("fingerprint", "ts")
+	} else {
+		sb.OrderBy("fingerprint", "unix_milli")
+	}
+
+	query, args := sb.BuildWithFlavor(sqlbuilder.ClickHouse)
+	return query, args, nil
+}
+
+// applySeriesConditions adds the WHERE conditions of a series table scan for
+// the given matchers and window. __name__ matchers translate to the
+// metric_name column (all four matcher types — the v1 client silently
+// returned nothing for regex metric names); every other matcher translates
+// to a JSONExtractString condition on the labels column. An equality matcher
+// against "" matches series without the label, mirroring PromQL, because
+// JSONExtractString returns "" for missing keys. Regexes are anchored:
+// PromQL matchers match the whole value, while ClickHouse match() searches
+// for a partial match — without anchoring, =~"api" would also select
+// "x-api-y".
+func applySeriesConditions(sb *sqlbuilder.SelectBuilder, start, end int64, matchers []*labels.Matcher) error {
+	for _, m := range matchers {
+		if m.Name != metricNameLabel {
+			continue
+		}
+		switch m.Type {
+		case labels.MatchEqual:
+			sb.Where(sb.EQ("metric_name", m.Value))
+		case labels.MatchNotEqual:
+			sb.Where(sb.NE("metric_name", m.Value))
+		case labels.MatchRegexp:
+			sb.Where(fmt.Sprintf("match(metric_name, %s)", sb.Var(anchorRegex(m.Value))))
+		case labels.MatchNotRegexp:
+			sb.Where(fmt.Sprintf("NOT match(metric_name, %s)", sb.Var(anchorRegex(m.Value))))
+		default:
+			return errors.NewInvalidInputf(errors.CodeInvalidInput, "unsupported matcher type %q for __name__", m.Type)
+		}
+	}
+
+	sb.Where("temporality IN ['Cumulative', 'Unspecified']")
+	sb.Where(fmt.Sprintf("__normalized = %v", !constants.IsDotMetricsEnabled))
+	sb.Where(sb.GTE("unix_milli", start), sb.LT("unix_milli", end))
+
+	for _, m := range matchers {
+		if m.Name == metricNameLabel {
+			continue
+		}
+		switch m.Type {
+		case labels.MatchEqual:
+			sb.Where(fmt.Sprintf("JSONExtractString(labels, %s) = %s", sb.Var(m.Name), sb.Var(m.Value)))
+		case labels.MatchNotEqual:
+			sb.Where(fmt.Sprintf("JSONExtractString(labels, %s) != %s", sb.Var(m.Name), sb.Var(m.Value)))
+		case labels.MatchRegexp:
+			sb.Where(fmt.Sprintf("match(JSONExtractString(labels, %s), %s)", sb.Var(m.Name), sb.Var(anchorRegex(m.Value))))
+		case labels.MatchNotRegexp:
+			sb.Where(fmt.Sprintf("NOT match(JSONExtractString(labels, %s), %s)", sb.Var(m.Name), sb.Var(anchorRegex(m.Value))))
+		default:
+			return errors.NewInvalidInputf(errors.CodeInvalidInput, "unsupported matcher type %q", m.Type)
+		}
+	}
+
+	return nil
+}
+
+// anchorRegex turns a PromQL regex into its fully-anchored form (see
+// applySeriesConditions).
+func anchorRegex(v string) string {
+	return "^(?:" + v + ")$"
+}
+
+// inlineFingerprintFilter renders "IN (fp1, fp2, ...)" with literal uint64s.
+func inlineFingerprintFilter(fingerprints []uint64) string {
+	var b strings.Builder
+	b.Grow(len(fingerprints)*21 + 8)
+	b.WriteString("IN (")
+	for i, fp := range fingerprints {
+		if i > 0 {
+			b.WriteString(", ")
+		}
+		b.WriteString(strconv.FormatUint(fp, 10))
+	}
+	b.WriteString(")")
+	return b.String()
+}
+
+// lastSamplePerStep reduces an instant-selector fetch to the last sample of
+// each step bucket. Buckets are anchored at the selector's first evaluation
+// timestamp so that every bucket boundary coincides with an evaluation
+// timestamp: bucket 0 is (start, firstEval] (the initial lookback window)
+// and bucket i is (firstEval+(i-1)·step, firstEval+i·step]. Keeping only the
+// last sample per bucket is lossless: the engine resolves each evaluation
+// timestamp t to the latest sample in (t-lookback, t], and a non-final
+// sample of a bucket can never be that latest sample for any t on the
+// evaluation grid. Real timestamps are preserved, so the engine's own
+// lookback and staleness handling remain exact.
+type lastSamplePerStep struct {
+	firstEvalMs int64
+	stepMs      int64
+}
+
+func (t *lastSamplePerStep) bucketExpr() string {
+	if t.stepMs <= 0 {
+		// Instant query: a single evaluation at firstEval; one bucket.
+		return ""
+	}
+	return fmt.Sprintf(
+		"if(unix_milli <= %d, 0, intDiv(unix_milli - %d - 1, %d) + 1)",
+		t.firstEvalMs, t.firstEvalMs, t.stepMs,
+	)
+}

pkg/prometheus/clickhouseprometheusv2/sql_test.go

@@ -0,0 +1,148 @@
+package clickhouseprometheusv2
+
+import (
+	"testing"
+	"time"
+
+	"github.com/prometheus/prometheus/model/labels"
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+)
+
+func mustMatcher(t *testing.T, mt labels.MatchType, name, value string) *labels.Matcher {
+	t.Helper()
+	m, err := labels.NewMatcher(mt, name, value)
+	require.NoError(t, err)
+	return m
+}
+
+func TestTimeSeriesTableFor(t *testing.T) {
+	base := time.Date(2026, 7, 10, 3, 27, 0, 0, time.UTC).UnixMilli()
+
+	tests := []struct {
+		name      string
+		span      time.Duration
+		wantTable string
+		roundTo   time.Duration
+	}{
+		{"under 6h uses hourly table", 2 * time.Hour, distributedTimeSeriesV4, time.Hour},
+		{"under 1d uses 6h table", 12 * time.Hour, distributedTimeSeriesV46hrs, 6 * time.Hour},
+		{"under 1w uses 1d table", 3 * 24 * time.Hour, distributedTimeSeriesV41day, 24 * time.Hour},
+		{"over 1w uses 1w table", 10 * 24 * time.Hour, distributedTimeSeriesV41week, 7 * 24 * time.Hour},
+	}
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			start, table := timeSeriesTableFor(base, base+tt.span.Milliseconds())
+			assert.Equal(t, tt.wantTable, table)
+			assert.Zero(t, start%tt.roundTo.Milliseconds())
+			assert.LessOrEqual(t, start, base)
+		})
+	}
+}
+
+func TestBuildSeriesQuery(t *testing.T) {
+	start := int64(1_700_000_000_000)
+	end := start + time.Hour.Milliseconds()
+	// The series table window rounds down to the table's bucket boundary.
+	adjustedStart := start - (start % time.Hour.Milliseconds())
+
+	t.Run("equality name and label matchers", func(t *testing.T) {
+		query, args, err := buildSeriesQuery(start, end, []*labels.Matcher{
+			mustMatcher(t, labels.MatchEqual, "__name__", "http_requests_total"),
+			mustMatcher(t, labels.MatchEqual, "job", "api"),
+		})
+		require.NoError(t, err)
+		assert.Equal(t,
+			"SELECT fingerprint, any(labels) FROM signoz_metrics.distributed_time_series_v4 WHERE metric_name = ? AND temporality IN ['Cumulative', 'Unspecified'] AND __normalized = false AND unix_milli >= ? AND unix_milli < ? AND JSONExtractString(labels, ?) = ? GROUP BY fingerprint",
+			query,
+		)
+		assert.Equal(t, []any{"http_requests_total", adjustedStart, end, "job", "api"}, args)
+	})
+
+	t.Run("regex matchers are anchored", func(t *testing.T) {
+		_, args, err := buildSeriesQuery(start, end, []*labels.Matcher{
+			mustMatcher(t, labels.MatchEqual, "__name__", "up"),
+			mustMatcher(t, labels.MatchRegexp, "instance", "prod.*"),
+			mustMatcher(t, labels.MatchNotRegexp, "env", "dev|test"),
+		})
+		require.NoError(t, err)
+		assert.Equal(t, []any{"up", adjustedStart, end, "instance", "^(?:prod.*)$", "env", "^(?:dev|test)$"}, args)
+	})
+
+	t.Run("regex name matcher uses metric_name column", func(t *testing.T) {
+		query, args, err := buildSeriesQuery(start, end, []*labels.Matcher{
+			mustMatcher(t, labels.MatchRegexp, "__name__", "node_cpu.*|node_memory.*"),
+		})
+		require.NoError(t, err)
+		assert.Contains(t, query, "match(metric_name, ?)")
+		assert.NotContains(t, query, "JSONExtractString")
+		assert.Equal(t, []any{"^(?:node_cpu.*|node_memory.*)$", adjustedStart, end}, args)
+	})
+
+	t.Run("no name matcher omits metric_name condition", func(t *testing.T) {
+		query, _, err := buildSeriesQuery(start, end, []*labels.Matcher{
+			mustMatcher(t, labels.MatchEqual, "job", "api"),
+		})
+		require.NoError(t, err)
+		assert.NotContains(t, query, "metric_name")
+	})
+}
+
+func TestBuildSamplesQuery(t *testing.T) {
+	start := int64(1_700_000_000_000)
+	end := start + time.Hour.Milliseconds()
+	adjustedStart := start - (start % time.Hour.Milliseconds())
+	matchers := []*labels.Matcher{
+		mustMatcher(t, labels.MatchEqual, "__name__", "up"),
+		mustMatcher(t, labels.MatchEqual, "job", "api"),
+	}
+
+	t.Run("raw with inline fingerprints", func(t *testing.T) {
+		query, args, err := buildSamplesQuery(start, end, []string{"up"}, []uint64{7, 42}, matchers, nil)
+		require.NoError(t, err)
+		assert.Equal(t,
+			"SELECT fingerprint, unix_milli, value, flags FROM signoz_metrics.distributed_samples_v4 WHERE metric_name = ? AND temporality IN ['Cumulative', 'Unspecified'] AND fingerprint IN (7, 42) AND unix_milli >= ? AND unix_milli <= ? ORDER BY fingerprint, unix_milli",
+			query,
+		)
+		assert.Equal(t, []any{"up", start, end}, args)
+	})
+
+	t.Run("last-sample-per-step groups by step bucket anchored at first eval", func(t *testing.T) {
+		lastPerStep := &lastSamplePerStep{firstEvalMs: start + 299_999, stepMs: 60_000}
+		query, _, err := buildSamplesQuery(start, end, []string{"up"}, []uint64{7}, matchers, lastPerStep)
+		require.NoError(t, err)
+		assert.Contains(t, query, "argMax(value, unix_milli) AS val")
+		assert.Contains(t, query, "argMax(flags, unix_milli) AS fl")
+		assert.Contains(t, query, "GROUP BY fingerprint, if(unix_milli <= 1700000299999, 0, intDiv(unix_milli - 1700000299999 - 1, 60000) + 1)")
+		assert.Contains(t, query, "ORDER BY fingerprint, ts")
+		// Aliases must not shadow the source columns referenced in WHERE.
+		assert.NotContains(t, query, "AS unix_milli")
+		assert.NotContains(t, query, "AS value")
+		assert.NotContains(t, query, "AS flags")
+	})
+
+	t.Run("instant query keeps one bucket", func(t *testing.T) {
+		lastPerStep := &lastSamplePerStep{firstEvalMs: end, stepMs: 0}
+		query, _, err := buildSamplesQuery(start, end, []string{"up"}, []uint64{7}, matchers, lastPerStep)
+		require.NoError(t, err)
+		assert.Contains(t, query, "GROUP BY fingerprint ORDER BY fingerprint, ts")
+		assert.NotContains(t, query, "intDiv")
+	})
+
+	t.Run("over-limit set becomes a shard-local semi-join", func(t *testing.T) {
+		query, args, err := buildSamplesQuery(start, end, []string{"up"}, nil, matchers, nil)
+		require.NoError(t, err)
+		assert.Contains(t, query, "fingerprint IN (SELECT fingerprint FROM signoz_metrics.time_series_v4 WHERE ")
+		assert.NotContains(t, query, "GLOBAL IN")
+		// Args follow placeholder order: samples metric name, the semi-join's
+		// series predicates, then the samples window bounds.
+		assert.Equal(t, []any{"up", "up", adjustedStart, end, "job", "api", start, end}, args)
+	})
+
+	t.Run("multiple metric names from regex selector", func(t *testing.T) {
+		query, args, err := buildSamplesQuery(start, end, []string{"node_cpu", "node_memory"}, []uint64{7}, matchers, nil)
+		require.NoError(t, err)
+		assert.Contains(t, query, "metric_name IN (?, ?)")
+		assert.Equal(t, []any{"node_cpu", "node_memory", start, end}, args)
+	})
+}

pkg/prometheus/clickhouseprometheusv2/tables.go

@@ -0,0 +1,65 @@
+package clickhouseprometheusv2
+
+import "time"
+
+const (
+	// metricNameLabel is the reserved PromQL label holding the metric name.
+	metricNameLabel string = "__name__"
+
+	databaseName                 string = "signoz_metrics"
+	distributedTimeSeriesV4      string = "distributed_time_series_v4"
+	distributedTimeSeriesV46hrs  string = "distributed_time_series_v4_6hrs"
+	distributedTimeSeriesV41day  string = "distributed_time_series_v4_1day"
+	distributedTimeSeriesV41week string = "distributed_time_series_v4_1week"
+	distributedSamplesV4         string = "distributed_samples_v4"
+
+	localTimeSeriesV4      string = "time_series_v4"
+	localTimeSeriesV46hrs  string = "time_series_v4_6hrs"
+	localTimeSeriesV41day  string = "time_series_v4_1day"
+	localTimeSeriesV41week string = "time_series_v4_1week"
+)
+
+// localTimeSeriesTable maps a distributed time series table to its shard-local
+// table. Samples and time series shard on the same key
+// (cityHash64(env, temporality, metric_name, fingerprint)), so a query whose
+// top-level FROM is the distributed samples table can join or semi-join the
+// local time series table inside each shard: the shard rewrite runs the
+// subquery against the shard's own series rows, which are exactly the series
+// of the shard's samples. No broadcast, no initiator-side join.
+func localTimeSeriesTable(distributed string) string {
+	switch distributed {
+	case distributedTimeSeriesV46hrs:
+		return localTimeSeriesV46hrs
+	case distributedTimeSeriesV41day:
+		return localTimeSeriesV41day
+	case distributedTimeSeriesV41week:
+		return localTimeSeriesV41week
+	default:
+		return localTimeSeriesV4
+	}
+}
+
+var (
+	oneHourInMilliseconds  = time.Hour.Milliseconds()
+	sixHoursInMilliseconds = time.Hour.Milliseconds() * 6
+	oneDayInMilliseconds   = time.Hour.Milliseconds() * 24
+	oneWeekInMilliseconds  = time.Hour.Milliseconds() * 24 * 7
+)
+
+// timeSeriesTableFor returns the adjusted start and the time series table for
+// the window. Time series tables hold one row per (fingerprint, bucket), with
+// bucket granularities of 1h, 6h, 1d and 1w; the start is rounded down to the
+// bucket boundary so a window beginning mid-bucket still matches the bucket's
+// row.
+func timeSeriesTableFor(start, end int64) (int64, string) {
+	switch {
+	case end-start < sixHoursInMilliseconds:
+		return start - (start % oneHourInMilliseconds), distributedTimeSeriesV4
+	case end-start < oneDayInMilliseconds:
+		return start - (start % sixHoursInMilliseconds), distributedTimeSeriesV46hrs
+	case end-start < oneWeekInMilliseconds:
+		return start - (start % oneDayInMilliseconds), distributedTimeSeriesV41day
+	default:
+		return start - (start % oneWeekInMilliseconds), distributedTimeSeriesV41week
+	}
+}

pkg/prometheus/clickhouseprometheusv2/transpiler.go

@@ -0,0 +1,485 @@
+package clickhouseprometheusv2
+
+import (
+	"fmt"
+	"strings"
+
+	"github.com/prometheus/prometheus/model/labels"
+	"github.com/prometheus/prometheus/promql/parser"
+)
+
+// The compiler turns PromQL subtrees into single ClickHouse queries built on
+// the timeSeries*ToGrid aggregate functions (CH >= 25.6), whose semantics
+// were verified against this repo's vendored engine: exact extrapolatedRate
+// behavior including counter resets, the counter zero-point clamp, the
+// 1.1x-average extrapolation threshold, left-open windows, the >= 2 samples
+// rule, stale-marker shadowing, and millisecond grid starts. Sample rows
+// never leave ClickHouse: one row per output series comes back, holding the
+// whole grid as an array.
+//
+// Scope (the allowlist): an optional sum/min/max/avg/count by/without
+// aggregation over a core unit — a rate/increase/delta/irate/idelta range
+// selection, an instant vector selection, or an avg/min/max/sum/count/last
+// _over_time window — plus number-literal arithmetic/comparisons and unary
+// minus on top. Units inside fixed-resolution subqueries evaluate on the
+// subquery's own grid. Everything else either falls back to the engine over
+// this package's querier, or — when a transpilable subtree sits under a
+// non-transpilable node — runs hybrid: the subtree's grids are computed in
+// ClickHouse and substituted into the engine as synthetic series (see
+// compiler_exec.go). See doc.go for the fallback list and the reasons behind
+// each entry.
+
+// rangeFn is a transpilable range-vector function.
+type rangeFn string
+
+const (
+	fnRate     rangeFn = "rate"
+	fnIncrease rangeFn = "increase"
+	fnDelta    rangeFn = "delta"
+	fnIRate    rangeFn = "irate"
+	fnIDelta   rangeFn = "idelta"
+)
+
+var gridFunction = map[rangeFn]string{
+	fnRate:     "timeSeriesRateToGrid",
+	fnIncrease: "timeSeriesRateToGrid", // increase == rate * range seconds, exactly (same factor algebra)
+	fnDelta:    "timeSeriesDeltaToGrid",
+	fnIRate:    "timeSeriesInstantRateToGrid",
+	fnIDelta:   "timeSeriesInstantDeltaToGrid",
+}
+
+// scalarOp is one number-literal arithmetic or comparison applied to a
+// compiled vector, evaluated in Go during assembly with the same float64
+// operations the engine uses.
+type scalarOp struct {
+	op           parser.ItemType
+	scalar       float64
+	scalarOnLeft bool
+	returnBool   bool
+}
+
+// isComparison reports whether the op is a filtering/bool comparison, which
+// preserves the metric name (arithmetic drops it).
+func (o scalarOp) isComparison() bool {
+	return o.op.IsComparisonOperator()
+}
+
+// unitKind is the selector shape at the bottom of a core unit.
+type unitKind int
+
+const (
+	// unitRange: rate/increase/delta/irate/idelta over a matrix selector.
+	unitRange unitKind = iota
+	// unitInstant: a plain vector selector resolved per grid point with
+	// lookback and stale-marker shadowing.
+	unitInstant
+	// unitOverTime: avg/min/max/sum/count/last_over_time over a matrix
+	// selector (aggregation over the window's samples, stale rows excluded).
+	unitOverTime
+)
+
+// coreUnit is one transpilable subtree: selector [-> range function] ->
+// optional aggregation -> scalar op pipeline.
+type coreUnit struct {
+	kind     unitKind
+	matchers []*labels.Matcher
+	offsetMs int64
+	fn       rangeFn // unitRange
+	overFn   string  // unitOverTime: avg|min|max|sum|count|last
+	rangeMs  int64   // unitRange/unitOverTime window
+
+	hasAgg   bool
+	aggOp    parser.ItemType // SUM MIN MAX AVG COUNT
+	by       bool
+	grouping []string
+
+	ops []scalarOp
+}
+
+// keepsName reports whether the unit's output series keep their real
+// __name__: bare/comparison-filtered instant selectors and last_over_time do
+// (it returns the raw sample, name included); range functions, the other
+// *_over_time functions, aggregations, arithmetic and bool comparisons all
+// drop it — a bool comparison returns 0/1, not the sample, so the engine
+// drops the name there too. Units that keep the name cannot be substituted
+// as synthetic series in hybrid plans — the synthetic name would replace
+// the real one — but transpile fine as full plans, where assembly emits the
+// real names.
+func (u *coreUnit) keepsName() bool {
+	nameKeepingSelector := u.kind == unitInstant || (u.kind == unitOverTime && u.overFn == "last")
+	if !nameKeepingSelector || u.hasAgg {
+		return false
+	}
+	for _, op := range u.ops {
+		if !op.isComparison() || op.returnBool {
+			return false
+		}
+	}
+	return true
+}
+
+// gridContext is the evaluation grid a unit computes on. The query grid for
+// top-level units; for units inside subqueries, the subquery's own grid:
+// epoch-aligned multiples of its resolution covering the subquery window,
+// exactly as the engine derives it (engine.go, *parser.SubqueryExpr case).
+type gridContext struct {
+	startMs int64
+	endMs   int64
+	stepMs  int64
+}
+
+// subqueryGrid derives the inner grid for a subquery evaluated on outer:
+// interval S, end = outer end − offset, start = first multiple of S strictly
+// greater than outer start − offset − range.
+func subqueryGrid(outer gridContext, rangeMs, stepMs, offsetMs int64) gridContext {
+	lower := outer.startMs - offsetMs - rangeMs
+	start := stepMs * (lower / stepMs)
+	if start <= lower {
+		start += stepMs
+	}
+	return gridContext{startMs: start, endMs: outer.endMs - offsetMs, stepMs: stepMs}
+}
+
+// transpiledUnit is a coreUnit scheduled for execution, named for hybrid
+// substitution, carrying the grid it evaluates on.
+type transpiledUnit struct {
+	core coreUnit
+	name string // __signoz_transpiled_<n>__
+	grid gridContext
+}
+
+// transpilePlan is the outcome of classifying a query.
+type transpilePlan struct {
+	units []*transpiledUnit
+	grid  gridContext // the query's top-level grid
+	// full is set when the entire query is units[0]; otherwise rewritten
+	// holds the query with each unit replaced by a synthetic selector, to be
+	// evaluated by the engine over a hybrid storage.
+	full      bool
+	rewritten string
+}
+
+const syntheticNamePrefix = "__signoz_transpiled_"
+
+func syntheticName(i int) string {
+	return fmt.Sprintf("%s%d__", syntheticNamePrefix, i)
+}
+
+// classifyCore matches a subtree against the transpilable core shape.
+// stepMs gates second-granularity: the grid functions take whole-second step
+// and window parameters (grid *starts* are millisecond-precise).
+func classifyCore(node parser.Expr, stepMs int64) (*coreUnit, bool) {
+	unit := &coreUnit{}
+
+	expr := node
+	// Peel scalar ops and parens off the top, outermost first; ops apply in
+	// evaluation order, so prepend while peeling.
+	for {
+		switch n := expr.(type) {
+		case *parser.ParenExpr:
+			expr = n.Expr
+			continue
+		case *parser.UnaryExpr:
+			if n.Op != parser.SUB {
+				expr = n.Expr // unary '+' is a no-op
+				continue
+			}
+			// -x == -1 * x for every float64 (incl. NaN and signed zero).
+			unit.ops = append([]scalarOp{{op: parser.MUL, scalar: -1}}, unit.ops...)
+			expr = n.Expr
+			continue
+		case *parser.StepInvariantExpr:
+			// @-pinned expressions evaluate on a different grid.
+			return nil, false
+		case *parser.BinaryExpr:
+			lit, litOnLeft, ok := numberLiteralSide(n)
+			if !ok {
+				return nil, false
+			}
+			if !n.Op.IsOperator() && !n.Op.IsComparisonOperator() {
+				return nil, false
+			}
+			if n.Op == parser.ATAN2 {
+				// atan2 is arithmetic in PromQL but rarely used; keep the
+				// allowlist tight.
+				return nil, false
+			}
+			returnBool := n.ReturnBool
+			unit.ops = append([]scalarOp{{op: n.Op, scalar: lit, scalarOnLeft: litOnLeft, returnBool: returnBool}}, unit.ops...)
+			if litOnLeft {
+				expr = n.RHS
+			} else {
+				expr = n.LHS
+			}
+			continue
+		}
+		break
+	}
+
+	// Optional aggregation.
+	if agg, ok := expr.(*parser.AggregateExpr); ok {
+		switch agg.Op {
+		case parser.SUM, parser.MIN, parser.MAX, parser.AVG, parser.COUNT:
+		default:
+			return nil, false
+		}
+		for _, g := range agg.Grouping {
+			if g == metricNameLabel {
+				// by(__name__)/without(__name__) over synthetic or compiled
+				// output needs name bookkeeping the compiler doesn't do.
+				return nil, false
+			}
+		}
+		unit.hasAgg = true
+		unit.aggOp = agg.Op
+		unit.by = !agg.Without
+		unit.grouping = agg.Grouping
+		expr = agg.Expr
+		for {
+			if p, ok := expr.(*parser.ParenExpr); ok {
+				expr = p.Expr
+				continue
+			}
+			break
+		}
+	}
+
+	// The grid functions take whole-second steps; stepMs == 0 is an instant
+	// query (single-point grid).
+	if stepMs < 0 || stepMs%1000 != 0 {
+		return nil, false
+	}
+
+	// Bare instant selector: resolved per grid point with lookback and
+	// stale-marker shadowing (see compiler_sql.go).
+	if vs, ok := expr.(*parser.VectorSelector); ok {
+		if vs.Timestamp != nil || vs.StartOrEnd != 0 || vs.Anchored || vs.Smoothed {
+			return nil, false
+		}
+		offsetMs := vs.OriginalOffset.Milliseconds()
+		if offsetMs < 0 {
+			return nil, false
+		}
+		unit.kind = unitInstant
+		unit.offsetMs = offsetMs
+		unit.matchers = vs.LabelMatchers
+		return unit, true
+	}
+
+	// Range or *_over_time function over a plain matrix selector.
+	call, ok := expr.(*parser.Call)
+	if !ok {
+		return nil, false
+	}
+	var fn rangeFn
+	var overFn string
+	switch call.Func.Name {
+	case "rate":
+		fn = fnRate
+	case "increase":
+		fn = fnIncrease
+	case "delta":
+		fn = fnDelta
+	case "irate":
+		fn = fnIRate
+	case "idelta":
+		fn = fnIDelta
+	case "avg_over_time", "min_over_time", "max_over_time", "sum_over_time", "count_over_time", "last_over_time":
+		overFn = strings.TrimSuffix(call.Func.Name, "_over_time")
+	default:
+		return nil, false
+	}
+	if len(call.Args) != 1 {
+		return nil, false
+	}
+	ms, ok := call.Args[0].(*parser.MatrixSelector)
+	if !ok {
+		return nil, false
+	}
+	vs, ok := ms.VectorSelector.(*parser.VectorSelector)
+	if !ok {
+		return nil, false
+	}
+	if vs.Timestamp != nil || vs.StartOrEnd != 0 || vs.Anchored || vs.Smoothed {
+		return nil, false
+	}
+
+	rangeMs := ms.Range.Milliseconds()
+	offsetMs := vs.OriginalOffset.Milliseconds()
+	if rangeMs <= 0 || rangeMs%1000 != 0 || offsetMs < 0 {
+		return nil, false
+	}
+
+	if overFn != "" {
+		unit.kind = unitOverTime
+		unit.overFn = overFn
+	} else {
+		unit.kind = unitRange
+		unit.fn = fn
+	}
+	unit.rangeMs = rangeMs
+	unit.offsetMs = offsetMs
+	unit.matchers = vs.LabelMatchers
+	return unit, true
+}
+
+// numberLiteralSide returns the number literal on one side of a binary
+// expression (peeling parens and unary minus), and which side it is on.
+func numberLiteralSide(b *parser.BinaryExpr) (float64, bool, bool) {
+	if v, ok := literalValue(b.LHS); ok {
+		return v, true, true
+	}
+	if v, ok := literalValue(b.RHS); ok {
+		return v, false, true
+	}
+	return 0, false, false
+}
+
+func literalValue(e parser.Expr) (float64, bool) {
+	neg := false
+	for {
+		switch n := e.(type) {
+		case *parser.ParenExpr:
+			e = n.Expr
+			continue
+		case *parser.StepInvariantExpr:
+			e = n.Expr
+			continue
+		case *parser.UnaryExpr:
+			if n.Op == parser.SUB {
+				neg = !neg
+			}
+			e = n.Expr
+			continue
+		case *parser.NumberLiteral:
+			if neg {
+				return -n.Val, true
+			}
+			return n.Val, true
+		default:
+			return 0, false
+		}
+	}
+}
+
+// classify builds the compile plan for a query: full when the root is a core
+// unit, hybrid when core units sit strictly below the root (including inside
+// fixed-resolution subqueries, computed on the subquery grid), none
+// otherwise.
+func classify(root parser.Expr, grid gridContext) (*transpilePlan, bool) {
+	if unit, ok := classifyCore(root, grid.stepMs); ok {
+		return &transpilePlan{
+			units: []*transpiledUnit{{core: *unit, name: syntheticName(0), grid: grid}},
+			grid:  grid,
+			full:  true,
+		}, true
+	}
+
+	plan := &transpilePlan{grid: grid}
+	rewritten := rewrite(root, grid, plan, false)
+	if len(plan.units) == 0 {
+		return nil, false
+	}
+	plan.rewritten = rewritten.String()
+	return plan, true
+}
+
+// rewrite walks top-down replacing maximal transpilable subtrees with synthetic
+// vector selectors. nameSensitive marks scopes where an ancestor's semantics
+// depend on __name__ (grouping or vector matching on it): synthetic series
+// carry a synthetic __name__, so substitution there would change results.
+// Fixed-resolution subqueries recurse with the subquery's own grid; scopes
+// whose evaluation grid is unknowable (@-pinned, default-resolution
+// subqueries) are not entered.
+func rewrite(node parser.Expr, grid gridContext, plan *transpilePlan, nameSensitive bool) parser.Expr {
+	if node == nil {
+		return nil
+	}
+
+	if !nameSensitive {
+		// Units whose output keeps the real __name__ (bare instant selectors)
+		// cannot be substituted: the synthetic name would replace it in the
+		// engine's output. They still compile as full plans.
+		if unit, ok := classifyCore(node, grid.stepMs); ok && !unit.keepsName() {
+			cu := &transpiledUnit{core: *unit, name: syntheticName(len(plan.units)), grid: grid}
+			plan.units = append(plan.units, cu)
+			return &parser.VectorSelector{
+				Name: cu.name,
+				LabelMatchers: []*labels.Matcher{
+					labels.MustNewMatcher(labels.MatchEqual, metricNameLabel, cu.name),
+				},
+				PosRange: node.PositionRange(),
+			}
+		}
+	}
+
+	switch n := node.(type) {
+	case *parser.ParenExpr:
+		n.Expr = rewrite(n.Expr, grid, plan, nameSensitive)
+	case *parser.UnaryExpr:
+		n.Expr = rewrite(n.Expr, grid, plan, nameSensitive)
+	case *parser.AggregateExpr:
+		sensitive := nameSensitive || groupingUsesName(n.Grouping)
+		n.Expr = rewrite(n.Expr, grid, plan, sensitive)
+		// n.Param is a scalar/string; nothing transpilable inside for our core.
+	case *parser.Call:
+		for i, arg := range n.Args {
+			n.Args[i] = rewrite(arg, grid, plan, nameSensitive)
+		}
+	case *parser.BinaryExpr:
+		sensitive := nameSensitive || vectorMatchingUsesName(n.VectorMatching)
+		n.LHS = rewrite(n.LHS, grid, plan, sensitive)
+		n.RHS = rewrite(n.RHS, grid, plan, sensitive)
+	case *parser.SubqueryExpr:
+		// The alert-smoothing idiom fn_over_time((expr)[R:S]) dominates real
+		// rule fleets; inner units evaluate on the subquery grid, and the
+		// engine does the smoothing over the synthetic series. Requires an
+		// explicit whole-second resolution (S == 0 needs the engine's
+		// default-interval function) and no @ pinning.
+		stepMs := n.Step.Milliseconds()
+		rangeMs := n.Range.Milliseconds()
+		offsetMs := n.OriginalOffset.Milliseconds()
+		if n.Timestamp == nil && n.StartOrEnd == 0 &&
+			stepMs > 0 && stepMs%1000 == 0 && rangeMs%1000 == 0 && offsetMs >= 0 {
+			inner := subqueryGrid(grid, rangeMs, stepMs, offsetMs)
+			n.Expr = rewrite(n.Expr, inner, plan, nameSensitive)
+		}
+	case *parser.StepInvariantExpr, *parser.MatrixSelector,
+		*parser.VectorSelector, *parser.NumberLiteral, *parser.StringLiteral:
+		// Leaves, or scopes substitution must not enter.
+	}
+	return node
+}
+
+func groupingUsesName(grouping []string) bool {
+	for _, g := range grouping {
+		if g == metricNameLabel {
+			return true
+		}
+	}
+	return false
+}
+
+func vectorMatchingUsesName(vm *parser.VectorMatching) bool {
+	if vm == nil {
+		return false
+	}
+	for _, l := range append(append([]string{}, vm.MatchingLabels...), vm.Include...) {
+		if l == metricNameLabel {
+			return true
+		}
+	}
+	// Default (all-labels) matching ignores __name__, and by()/ignoring()
+	// lists were checked above.
+	return false
+}
+
+// isSyntheticSelector reports whether matchers target a compiled unit.
+func isSyntheticSelector(matchers []*labels.Matcher) (string, bool) {
+	for _, m := range matchers {
+		if m.Name == metricNameLabel && m.Type == labels.MatchEqual && strings.HasPrefix(m.Value, syntheticNamePrefix) {
+			return m.Value, true
+		}
+	}
+	return "", false
+}

pkg/prometheus/clickhouseprometheusv2/transpiler_corpus_test.go

@@ -0,0 +1,150 @@
+package clickhouseprometheusv2
+
+import (
+	"bufio"
+	"encoding/json"
+	"fmt"
+	"os"
+	"regexp"
+	"sort"
+	"strings"
+	"testing"
+
+	"github.com/prometheus/prometheus/promql/parser"
+	"github.com/stretchr/testify/require"
+)
+
+// TestClassifyCorpus measures real-workload compiler coverage: it classifies
+// every query of a JSON-lines corpus (one JSON-encoded PromQL string per
+// line) with the live classifier and reports full / hybrid / fallback
+// shares. Skipped unless PROMQL_CORPUS points to one or more files
+// (comma-separated). Dashboard template variables are substituted with
+// placeholder values before parsing, mirroring the production render step.
+//
+//	PROMQL_CORPUS=corpus-a.jsonl,corpus-b.jsonl go test -run TestClassifyCorpus -v
+func TestClassifyCorpus(t *testing.T) {
+	corpus := os.Getenv("PROMQL_CORPUS")
+	if corpus == "" {
+		t.Skip("PROMQL_CORPUS not set")
+	}
+
+	varRe := regexp.MustCompile(`\{\{\s*\.?[\w.]+\s*\}\}|\[\[\s*[\w.]+\s*\]\]|\$[\w.]+`)
+	promParser := parser.NewParser(parser.Options{})
+
+	for _, path := range strings.Split(corpus, ",") {
+		f, err := os.Open(path)
+		require.NoError(t, err)
+
+		var full, hybrid, fallbackInstant, fallbackOther, parseErrs int
+		fallbackReasons := map[string]int{}
+
+		scanner := bufio.NewScanner(f)
+		scanner.Buffer(make([]byte, 1024*1024), 1024*1024)
+		for scanner.Scan() {
+			var query string
+			require.NoError(t, json.Unmarshal(scanner.Bytes(), &query))
+			query = varRe.ReplaceAllString(query, "placeholder")
+
+			expr, err := promParser.ParseExpr(query)
+			if err != nil {
+				parseErrs++
+				continue
+			}
+
+			plan, ok := classify(expr, gridContext{startMs: 1_700_000_000_000, endMs: 1_700_007_200_000, stepMs: 60_000})
+			switch {
+			case ok && plan.full:
+				full++
+			case ok:
+				hybrid++
+			default:
+				reason := fallbackShape(expr)
+				fallbackReasons[reason]++
+				if reason == "instant-selector shape (last-sample-per-step engine path)" {
+					fallbackInstant++
+				} else {
+					fallbackOther++
+				}
+			}
+		}
+		require.NoError(t, scanner.Err())
+		_ = f.Close()
+
+		total := full + hybrid + fallbackInstant + fallbackOther
+		if total == 0 {
+			t.Logf("%s: no parseable queries (%d parse errors)", path, parseErrs)
+			continue
+		}
+		t.Logf("%s: %d queries — full=%d (%.0f%%) hybrid=%d (%.0f%%) fallback=%d (%.0f%%; instant-shape=%d) parse_errors=%d",
+			path, total,
+			full, 100*float64(full)/float64(total),
+			hybrid, 100*float64(hybrid)/float64(total),
+			fallbackInstant+fallbackOther, 100*float64(fallbackInstant+fallbackOther)/float64(total),
+			fallbackInstant, parseErrs)
+
+		reasons := make([]string, 0, len(fallbackReasons))
+		for r := range fallbackReasons {
+			reasons = append(reasons, r)
+		}
+		sort.Slice(reasons, func(i, j int) bool { return fallbackReasons[reasons[i]] > fallbackReasons[reasons[j]] })
+		for _, r := range reasons {
+			t.Logf("  fallback %4d  %s", fallbackReasons[r], r)
+		}
+	}
+}
+
+// fallbackShape buckets a non-transpilable query by why it stays on the engine
+// path, to separate "already served well" (instant selectors on the last-sample-per-step
+// path) from genuine compiler gaps.
+func fallbackShape(expr parser.Expr) string {
+	var hasMatrix, hasSubquery, hasAt, overTime bool
+	rangeFns := map[string]bool{"rate": true, "increase": true, "delta": true, "irate": true, "idelta": true}
+	var unsupportedFns []string
+	parser.Inspect(expr, func(node parser.Node, _ []parser.Node) error {
+		switch n := node.(type) {
+		case *parser.MatrixSelector:
+			hasMatrix = true
+		case *parser.SubqueryExpr:
+			hasSubquery = true
+		case *parser.VectorSelector:
+			if n.Timestamp != nil || n.StartOrEnd != 0 {
+				hasAt = true
+			}
+		case *parser.Call:
+			if strings.HasSuffix(n.Func.Name, "_over_time") {
+				overTime = true
+			} else if !rangeFns[n.Func.Name] {
+				unsupportedFns = append(unsupportedFns, n.Func.Name)
+			}
+		}
+		return nil
+	})
+
+	switch {
+	case hasSubquery:
+		return "subquery"
+	case hasAt:
+		return "@ modifier"
+	case overTime:
+		return "*_over_time range function"
+	case !hasMatrix:
+		return "instant-selector shape (last-sample-per-step engine path)"
+	case len(unsupportedFns) > 0:
+		return fmt.Sprintf("range shape with unsupported function(s): %s", strings.Join(dedupe(unsupportedFns), ",")) //nolint:makezero
+	default:
+		return "other range shape"
+	}
+}
+
+func dedupe(in []string) []string {
+	seen := map[string]bool{}
+	var out []string
+	for _, s := range in {
+		if !seen[s] {
+			seen[s] = true
+			out = append(out, s)
+		}
+	}
+	sort.Strings(out)
+	return out
+}

pkg/prometheus/clickhouseprometheusv2/transpiler_exec.go

@@ -0,0 +1,463 @@
+package clickhouseprometheusv2
+
+import (
+	"context"
+	"encoding/json"
+	"math"
+	"sort"
+	"sync/atomic"
+	"time"
+
+	"github.com/SigNoz/signoz/pkg/errors"
+	"github.com/SigNoz/signoz/pkg/prometheus"
+	"github.com/prometheus/prometheus/model/labels"
+	promValue "github.com/prometheus/prometheus/model/value"
+	"github.com/prometheus/prometheus/promql"
+	"github.com/prometheus/prometheus/promql/parser"
+	"github.com/prometheus/prometheus/storage"
+	"golang.org/x/sync/errgroup"
+)
+
+// executor evaluates transpilable PromQL directly in ClickHouse, falling
+// back (ok=false) whenever the query shape or the step doesn't qualify. The
+// timeSeries*ToGrid functions it builds on are assumed available: the
+// supported ClickHouse floor is >= 25.6.
+type executor struct {
+	client *client
+	engine *prometheus.Engine
+	parser prometheus.Parser
+}
+
+// TryExecuteRange transpiles and runs the query in ClickHouse when its shape
+// is in the allowlist. ok=false means "not transpilable" and carries no
+// error; the caller runs the engine path.
+func (e *executor) TryExecuteRange(ctx context.Context, qs string, start, end time.Time, step time.Duration) (promql.Matrix, bool, error) {
+	expr, err := e.parser.ParseExpr(qs)
+	if err != nil {
+		// Let the engine path produce the (enhanced) parse error.
+		return nil, false, nil
+	}
+
+	plan, ok := classify(expr, queryGrid(start, end, step))
+	if !ok {
+		return nil, false, nil
+	}
+
+	// timeSeriesLastToGrid widens its window to max(window, step) — probed: a
+	// sample aged (window, step] still fills the slot — while the rate/delta
+	// family enforces the window strictly. The Last-style kinds therefore
+	// transpile only when their window covers the step; otherwise the engine
+	// path serves them exactly.
+	for _, unit := range plan.units {
+		lastStyle := unit.core.kind == unitInstant || (unit.core.kind == unitOverTime && unit.core.overFn == "last")
+		if !lastStyle {
+			continue
+		}
+		windowMs := unit.core.rangeMs
+		if unit.core.kind == unitInstant {
+			windowMs = e.client.lookbackMs
+		}
+		if windowMs < unit.grid.stepMs {
+			return nil, false, nil
+		}
+	}
+
+	// Evaluate every unit concurrently on its own grid (the query grid, or a
+	// subquery grid); each is one series lookup plus one grid query. The
+	// units share one grid-cell budget: transpiled results never pass
+	// through the engine's sample limiter, so without this a large
+	// series-count x grid-width query would buffer unbounded arrays — the
+	// OOM this provider exists to prevent.
+	results := make([][]transpiledSeries, len(plan.units))
+	var gridCells atomic.Int64
+	eg, egCtx := errgroup.WithContext(ctx)
+	for i, unit := range plan.units {
+		eg.Go(func() error {
+			res, err := e.executeUnit(egCtx, &unit.core, unit.grid, &gridCells)
+			if err != nil {
+				return err
+			}
+			results[i] = res
+			return nil
+		})
+	}
+	if err := eg.Wait(); err != nil {
+		return nil, true, err
+	}
+
+	if plan.full {
+		g := plan.units[0].grid
+		return toMatrix(results[0], g.startMs, g.stepMs), true, nil
+	}
+
+	matrix, err := e.executeHybrid(ctx, plan, results)
+	if err != nil {
+		return nil, true, err
+	}
+	return matrix, true, nil
+}
+
+// queryGrid derives the top-level evaluation grid; step 0 is an instant
+// query: a single evaluation at end, whatever start was.
+func queryGrid(start, end time.Time, step time.Duration) gridContext {
+	startMs, endMs, stepMs := start.UnixMilli(), end.UnixMilli(), step.Milliseconds()
+	if stepMs == 0 {
+		startMs = endMs
+	}
+	return gridContext{startMs: startMs, endMs: endMs, stepMs: stepMs}
+}
+
+// transpiledSeries is one output series of a unit: projected labels and one
+// value pointer per grid point (nil = absent).
+type transpiledSeries struct {
+	lset   labels.Labels
+	values []*float64
+}
+
+// executeUnit runs one core unit on its grid: series lookup (budgets,
+// fingerprints, metric names), then the single grid query, then the
+// scalar-op pipeline.
+func (e *executor) executeUnit(ctx context.Context, unit *coreUnit, grid gridContext, gridCells *atomic.Int64) ([]transpiledSeries, error) {
+	startMs, endMs, stepMs := grid.startMs, grid.endMs, grid.stepMs
+	windowMs := unit.rangeMs
+	if unit.kind == unitInstant {
+		windowMs = e.client.lookbackMs
+	}
+	dataStart := startMs - unit.offsetMs - windowMs
+	dataEnd := endMs - unit.offsetMs
+
+	seriesQuery, seriesArgs, err := buildSeriesQuery(dataStart, dataEnd, unit.matchers)
+	if err != nil {
+		return nil, err
+	}
+	lookup, err := e.client.selectSeries(ctx, seriesQuery, seriesArgs)
+	if err != nil {
+		return nil, err
+	}
+	if len(lookup.fingerprints) == 0 {
+		return nil, nil
+	}
+
+	// The result buffers one grid array per series; series count times grid
+	// width is the transpiled equivalent of fetched samples, counted before
+	// the arrays exist rather than after the memory is spent.
+	gridLen := int64(1)
+	if stepMs > 0 {
+		gridLen = (endMs-startMs)/stepMs + 1
+	}
+	if maxSamples := e.client.cfg.MaxFetchedSamples; maxSamples > 0 && gridCells.Add(int64(len(lookup.fingerprints))*gridLen) > maxSamples {
+		return nil, errors.NewInvalidInputf(
+			errors.CodeInvalidInput,
+			"promql query would buffer more than %d output points; narrow the selector or time range, or raise prometheus::clickhousev2::max_fetched_samples",
+			maxSamples,
+		)
+	}
+
+	query, args, err := buildUnitSQL(unit, lookup.metricNames, transpiledFingerprintFilter(lookup), dataStart, dataEnd, startMs, endMs, stepMs, e.client.lookbackMs)
+	if err != nil {
+		return nil, err
+	}
+
+	rows, err := e.client.telemetryStore.ClickhouseDB().Query(e.client.withContext(ctx, "transpiledUnit"), query, args...)
+	if err != nil {
+		return nil, err
+	}
+	defer rows.Close()
+
+	// Name-dropping units keep __name__ in the SQL group key so distinct
+	// metrics never merge server-side; the name comes off here, and a
+	// post-strip collision is the engine's duplicate-labelset error — v1
+	// would have errored, so silently inventing a merged series would be a
+	// divergence.
+	stripName := !unit.hasAgg && !unit.keepsName()
+	seen := make(map[uint64]string)
+
+	var out []transpiledSeries
+	var gkey string
+	var gridValues []*float64
+	for rows.Next() {
+		if err := rows.Scan(&gkey, &gridValues); err != nil {
+			return nil, err
+		}
+		lset, err := labelsFromGroupKey(gkey)
+		if err != nil {
+			return nil, err
+		}
+		if stripName {
+			name := lset.Get(metricNameLabel)
+			lset = labels.NewBuilder(lset).Del(metricNameLabel).Labels()
+			if prev, ok := seen[lset.Hash()]; ok && prev != name {
+				return nil, errors.NewInvalidInputf(errors.CodeInvalidInput, "vector cannot contain metrics with the same labelset")
+			}
+			seen[lset.Hash()] = name
+		}
+		values := make([]*float64, len(gridValues))
+		copy(values, gridValues)
+		applyScalarOps(unit.ops, values)
+		out = append(out, transpiledSeries{lset: lset, values: values})
+	}
+	if err := rows.Err(); err != nil {
+		return nil, err
+	}
+
+	sort.Slice(out, func(i, j int) bool { return labels.Compare(out[i].lset, out[j].lset) < 0 })
+	return out, nil
+}
+
+// transpiledFingerprintFilter returns the matched fingerprints as a sorted
+// slice when they fit the inline limit — literals engage the samples primary
+// key, and sorting keeps the statement deterministic for logging and tests.
+// Over the limit it returns nil: the unit query's INNER JOIN against the
+// local series subquery restricts to exactly the matched fingerprints
+// already, and a semi-join on the same predicates would only rescan the
+// series table.
+func transpiledFingerprintFilter(lookup *seriesLookup) []uint64 {
+	if len(lookup.fingerprints) > inlineFingerprintsLimit {
+		return nil
+	}
+	fingerprints := make([]uint64, 0, len(lookup.fingerprints))
+	for fp := range lookup.fingerprints {
+		fingerprints = append(fingerprints, fp)
+	}
+	sort.Slice(fingerprints, func(i, j int) bool { return fingerprints[i] < fingerprints[j] })
+	return fingerprints
+}
+
+// labelsFromGroupKey parses the toJSONString'd sorted [key, value] pairs.
+func labelsFromGroupKey(gkey string) (labels.Labels, error) {
+	var pairs [][]string
+	if err := json.Unmarshal([]byte(gkey), &pairs); err != nil {
+		return labels.EmptyLabels(), errors.WrapInternalf(err, errors.CodeInternal, "malformed compiled group key %q", gkey)
+	}
+	builder := labels.NewScratchBuilder(len(pairs))
+	for _, p := range pairs {
+		if len(p) != 2 {
+			return labels.EmptyLabels(), errors.NewInternalf(errors.CodeInternal, "malformed compiled group key pair %q", gkey)
+		}
+		builder.Add(p[0], p[1])
+	}
+	builder.Sort()
+	return builder.Labels(), nil
+}
+
+// applyScalarOps applies the number-literal op pipeline in place, with the
+// same float64 arithmetic and comparison-filter semantics as the engine.
+func applyScalarOps(ops []scalarOp, values []*float64) {
+	for _, op := range ops {
+		for i, v := range values {
+			if v == nil {
+				continue
+			}
+			lhs, rhs := *v, op.scalar
+			if op.scalarOnLeft {
+				lhs, rhs = op.scalar, *v
+			}
+			switch op.op {
+			case parser.ADD:
+				res := lhs + rhs
+				values[i] = &res
+			case parser.SUB:
+				res := lhs - rhs
+				values[i] = &res
+			case parser.MUL:
+				res := lhs * rhs
+				values[i] = &res
+			case parser.DIV:
+				res := lhs / rhs
+				values[i] = &res
+			case parser.MOD:
+				res := math.Mod(lhs, rhs)
+				values[i] = &res
+			case parser.POW:
+				res := math.Pow(lhs, rhs)
+				values[i] = &res
+			default:
+				keep := compare(op.op, lhs, rhs)
+				switch {
+				case op.returnBool:
+					res := 0.0
+					if keep {
+						res = 1.0
+					}
+					values[i] = &res
+				case keep:
+					// Filter comparisons keep the vector-side value.
+					vec := *v
+					values[i] = &vec
+				default:
+					values[i] = nil
+				}
+			}
+		}
+	}
+}
+
+func compare(op parser.ItemType, lhs, rhs float64) bool {
+	switch op {
+	case parser.EQLC:
+		return lhs == rhs
+	case parser.NEQ:
+		return lhs != rhs
+	case parser.GTR:
+		return lhs > rhs
+	case parser.LSS:
+		return lhs < rhs
+	case parser.GTE:
+		return lhs >= rhs
+	case parser.LTE:
+		return lhs <= rhs
+	}
+	return false
+}
+
+// toMatrix converts a unit result to a promql matrix on the query grid.
+func toMatrix(series []transpiledSeries, startMs, stepMs int64) promql.Matrix {
+	matrix := make(promql.Matrix, 0, len(series))
+	for _, s := range series {
+		var floats []promql.FPoint
+		for i, v := range s.values {
+			if v == nil {
+				continue
+			}
+			floats = append(floats, promql.FPoint{T: startMs + int64(i)*stepMs, F: *v})
+		}
+		if len(floats) == 0 {
+			continue
+		}
+		matrix = append(matrix, promql.Series{Metric: s.lset, Floats: floats})
+	}
+	return matrix
+}
+
+// executeHybrid substitutes each unit's grids into the engine as synthetic
+// series and evaluates the rewritten query over a storage that serves
+// synthetic selectors from memory and everything else from the live querier.
+// Absent grid points become stale markers so the engine's lookback cannot
+// resurrect the previous grid point. Each unit's synthetic samples sit on its
+// own grid (query grid, or subquery grid for units inside subqueries).
+func (e *executor) executeHybrid(ctx context.Context, plan *transpilePlan, results [][]transpiledSeries) (promql.Matrix, error) {
+	synthetic := make(map[string][]*series, len(plan.units))
+	staleMarker := math.Float64frombits(promValue.StaleNaN)
+
+	queryGrid := plan.grid
+
+	for i, unit := range plan.units {
+		g := unit.grid
+		gridLen := 1
+		if g.stepMs > 0 {
+			gridLen = int((g.endMs-g.startMs)/g.stepMs) + 1
+		}
+		list := make([]*series, 0, len(results[i]))
+		for _, cs := range results[i] {
+			builder := labels.NewBuilder(cs.lset)
+			builder.Set(metricNameLabel, unit.name)
+			s := &series{lset: builder.Labels()}
+			s.ts = make([]int64, 0, gridLen)
+			s.vs = make([]float64, 0, gridLen)
+			for idx := 0; idx < gridLen; idx++ {
+				t := g.startMs + int64(idx)*g.stepMs
+				var v float64
+				if idx < len(cs.values) && cs.values[idx] != nil {
+					v = *cs.values[idx]
+				} else {
+					v = staleMarker
+				}
+				s.ts = append(s.ts, t)
+				s.vs = append(s.vs, v)
+			}
+			list = append(list, s)
+		}
+		synthetic[unit.name] = list
+	}
+
+	hybrid := &hybridQueryable{client: e.client, synthetic: synthetic}
+
+	var qry promql.Query
+	var err error
+	if queryGrid.stepMs == 0 {
+		qry, err = e.engine.NewInstantQuery(ctx, hybrid, nil, plan.rewritten, time.UnixMilli(queryGrid.endMs))
+	} else {
+		qry, err = e.engine.NewRangeQuery(ctx, hybrid, nil, plan.rewritten, time.UnixMilli(queryGrid.startMs), time.UnixMilli(queryGrid.endMs), time.Duration(queryGrid.stepMs)*time.Millisecond)
+	}
+	if err != nil {
+		return nil, err
+	}
+	defer qry.Close()
+
+	res := qry.Exec(ctx)
+	if res.Err != nil {
+		return nil, res.Err
+	}
+
+	matrix, err := resultToMatrix(res)
+	if err != nil {
+		return nil, err
+	}
+
+	// Deep-copy before Close returns the result's slices to the engine pool,
+	// and drop the synthetic __name__ that filter comparisons preserve.
+	out := make(promql.Matrix, 0, len(matrix))
+	for _, s := range matrix {
+		lset := s.Metric
+		if name := lset.Get(metricNameLabel); len(name) >= len(syntheticNamePrefix) && name[:len(syntheticNamePrefix)] == syntheticNamePrefix {
+			builder := labels.NewBuilder(lset)
+			builder.Del(metricNameLabel)
+			lset = builder.Labels()
+		}
+		floats := make([]promql.FPoint, len(s.Floats))
+		copy(floats, s.Floats)
+		out = append(out, promql.Series{Metric: lset.Copy(), Floats: floats})
+	}
+	sort.Slice(out, func(i, j int) bool { return labels.Compare(out[i].Metric, out[j].Metric) < 0 })
+	return out, nil
+}
+
+func resultToMatrix(res *promql.Result) (promql.Matrix, error) {
+	switch v := res.Value.(type) {
+	case promql.Matrix:
+		return v, nil
+	case promql.Vector:
+		matrix := make(promql.Matrix, 0, len(v))
+		for _, s := range v {
+			matrix = append(matrix, promql.Series{Metric: s.Metric, Floats: []promql.FPoint{{T: s.T, F: s.F}}})
+		}
+		return matrix, nil
+	case promql.Scalar:
+		return promql.Matrix{{Metric: labels.EmptyLabels(), Floats: []promql.FPoint{{T: v.T, F: v.V}}}}, nil
+	default:
+		return nil, errors.NewInternalf(errors.CodeInternal, "unexpected hybrid result type %T", res.Value)
+	}
+}
+
+// hybridQueryable serves synthetic (compiled) selectors from memory and
+// everything else from the live storage.
+type hybridQueryable struct {
+	client    *client
+	synthetic map[string][]*series
+}
+
+func (h *hybridQueryable) Querier(mint, maxt int64) (storage.Querier, error) {
+	return &hybridQuerier{
+		querier:   querier{mint: mint, maxt: maxt, client: h.client},
+		synthetic: h.synthetic,
+	}, nil
+}
+
+type hybridQuerier struct {
+	querier
+	synthetic map[string][]*series
+}
+
+func (h *hybridQuerier) Select(ctx context.Context, sortSeries bool, hints *storage.SelectHints, matchers ...*labels.Matcher) storage.SeriesSet {
+	if name, ok := isSyntheticSelector(matchers); ok {
+		list := h.synthetic[name]
+		if sortSeries {
+			sorted := make([]*series, len(list))
+			copy(sorted, list)
+			sort.Slice(sorted, func(i, j int) bool { return labels.Compare(sorted[i].lset, sorted[j].lset) < 0 })
+			list = sorted
+		}
+		return newSeriesSet(list)
+	}
+	return h.querier.Select(ctx, sortSeries, hints, matchers...)
+}

pkg/prometheus/clickhouseprometheusv2/transpiler_sql.go

@@ -0,0 +1,303 @@
+package clickhouseprometheusv2
+
+import (
+	"fmt"
+	"strings"
+
+	"github.com/huandu/go-sqlbuilder"
+)
+
+// experimental gate for the timeSeries*ToGrid aggregate functions; attached
+// as a SETTINGS clause so telemetrystore hooks cannot clobber it.
+const gridFunctionsSetting = "SETTINGS allow_experimental_ts_to_grid_aggregate_function = 1"
+
+var aggForEach = map[string]string{
+	"sum":   "sumForEach",
+	"min":   "minForEach",
+	"max":   "maxForEach",
+	"avg":   "avgForEach",
+	"count": "countForEach",
+}
+
+// buildUnitSQL renders the single ClickHouse query evaluating a core unit
+// over the [startMs, endMs] / stepMs evaluation grid: per-series grids via a
+// timeSeries*ToGrid aggregate (or a windowed aggregation for *_over_time),
+// then spatial aggregation with -ForEach combinators grouped by a canonical
+// JSON key of the projected label pairs.
+//
+// The heavy level is shaped to run on the shards: the top-level FROM is the
+// distributed samples table and the group-key join partner is a subquery on
+// the shard-local time series table, so the shard rewrite executes the join
+// and the per-(fingerprint, gkey) grid aggregation next to the data —
+// complete by fingerprint co-locality (see localTimeSeriesTable) — and the
+// initiator only merges per-series grid states and applies the spatial
+// -ForEach step. Same layout as the telemetrymetrics statement builder. The
+// windowed *_over_time form is the exception: its ARRAY JOIN level
+// aggregates on the shards the same way, but the group-key join happens at
+// the initiator over the already-reduced per-(series, index) rows — pushing
+// it down would not move any data off the initiator (the reduced rows arrive
+// there either way), so the combined ARRAY JOIN + JOIN form buys nothing.
+//
+// inlineFingerprints carries the matched set when it fits the inline limit;
+// nil means over the limit, where the group-key join restricts on its own
+// (the windowed form, whose fan-out query has no join, falls back to a
+// shard-local semi-join so it does not expand every series of the metric).
+//
+// The selector's data window is offset-shifted; the resulting grid indices
+// map 1:1 onto the query grid (output ts = startMs + i*stepMs). Grid
+// parameters are rendered as literals — they are aggregate-function
+// parameters, not bindable values.
+//
+// Statements nest builder-rendered SQL as text, so the returned args must be
+// ordered by where each fragment lands in the final statement: ClickHouse
+// binds ? placeholders by position. A JOIN renders before WHERE, so a joined
+// subquery's args precede the outer query's own condition args.
+//
+// Row shape: (gkey String, grid Array(Nullable(Float64))). gkey is
+// toJSONString of the sorted projected [key, value] pairs; NULL grid points
+// are absent points (the engine's "no value here"), which the -ForEach
+// combinators preserve: an index where every series is NULL aggregates to
+// NULL, and countForEach's 0 is mapped back to NULL.
+func buildUnitSQL(unit *coreUnit, metricNames []string, inlineFingerprints []uint64, dataStart, dataEnd int64, startMs, endMs, stepMs, lookbackMs int64) (string, []any, error) {
+	selStart := startMs - unit.offsetMs
+	selEnd := endMs - unit.offsetMs
+	stepSec := stepMs / 1000
+	if stepSec == 0 {
+		// Instant query: start == end, so the grid has one point for any
+		// positive step.
+		stepSec = 1
+	}
+	windowMs := unit.rangeMs
+	if unit.kind == unitInstant {
+		windowMs = lookbackMs
+	}
+	windowSec := windowMs / 1000
+
+	adjustedTsStart, tsTable := timeSeriesTableFor(dataStart, dataEnd)
+
+	// seriesSub computes fingerprint -> group key. It reads the local series
+	// table when it rides inside the shard-rewritten samples query, and the
+	// distributed one when it joins at the initiator (windowed form).
+	seriesSub := func(table string) (string, []any, error) {
+		sub := sqlbuilder.NewSelectBuilder()
+		sub.Select("fingerprint", groupKeyExpr(unit)+" AS gkey")
+		sub.From(fmt.Sprintf("%s.%s", databaseName, table))
+		if err := applySeriesConditions(sub, adjustedTsStart, dataEnd, unit.matchers); err != nil {
+			return "", nil, err
+		}
+		sub.GroupBy("fingerprint", "gkey")
+		q, args := sub.BuildWithFlavor(sqlbuilder.ClickHouse)
+		return q, args, nil
+	}
+
+	// samplesConditions adds the samples-side WHERE. The samples table is
+	// aliased "points" in every kind: under the group-key join both sides
+	// carry a fingerprint column, so the filter must qualify it. A nil
+	// inline set adds no fingerprint condition — the join restricts.
+	samplesConditions := func(sb *sqlbuilder.SelectBuilder, excludeStale bool) {
+		switch len(metricNames) {
+		case 0:
+			// No name constraint derivable; correct but unable to use the
+			// metric_name primary-key prefix.
+		case 1:
+			sb.Where(sb.EQ("metric_name", metricNames[0]))
+		default:
+			sb.Where(sb.In("metric_name", sqlbuilder.List(metricNames)))
+		}
+		// temporality precedes metric_name in the samples primary key; the
+		// fingerprints already come from these temporalities, so this only
+		// helps granule pruning.
+		sb.Where("temporality IN ['Cumulative', 'Unspecified']")
+		if inlineFingerprints != nil {
+			sb.Where("points.fingerprint " + inlineFingerprintFilter(inlineFingerprints))
+		}
+		// Left-open window: a sample exactly at the window's lower boundary
+		// is never used (range selectors and lookback are both left-open).
+		sb.Where(sb.GT("unix_milli", selStart-windowMs), sb.LTE("unix_milli", selEnd))
+		if excludeStale {
+			// PromQL excludes stale markers from range vectors. Instant
+			// selectors need the stale rows for shadowing instead.
+			sb.Where("bitAnd(flags, 1) = 0")
+		}
+	}
+
+	// joinedInner builds the shard-side SELECT for the single-pass kinds:
+	// grid expression per (fingerprint, gkey), group-key join against the
+	// local series table.
+	joinedInner := func(gridExpr string, excludeStale bool) (string, []any, error) {
+		seriesSQL, seriesArgs, err := seriesSub(localTimeSeriesTable(tsTable))
+		if err != nil {
+			return "", nil, err
+		}
+		sb := sqlbuilder.NewSelectBuilder()
+		sb.Select("series.gkey AS gkey", gridExpr+" AS grid")
+		sb.From(fmt.Sprintf("%s.%s AS points", databaseName, distributedSamplesV4))
+		sb.JoinWithOption(sqlbuilder.InnerJoin, fmt.Sprintf("(%s) AS series", seriesSQL), "points.fingerprint = series.fingerprint")
+		samplesConditions(sb, excludeStale)
+		sb.GroupBy("points.fingerprint", "series.gkey")
+		q, args := sb.BuildWithFlavor(sqlbuilder.ClickHouse)
+		// The join text renders before WHERE: its args come first.
+		return q, append(seriesArgs, args...), nil
+	}
+
+	var inner string
+	var innerArgs []any
+	var err error
+	switch unit.kind {
+	case unitInstant:
+		// Instant selection with stale shadowing: the grid value is the last
+		// non-stale sample in (t-lookback, t], absent when the overall last
+		// sample in that window is a stale marker (verified semantics: the
+		// -If combinator applies to the grid aggregates, and NULL comparisons
+		// make a stale-latest point absent).
+		gridParams := fmt.Sprintf("(fromUnixTimestamp64Milli(%d), fromUnixTimestamp64Milli(%d), %d, %d)", selStart, selEnd, stepSec, windowSec)
+		gridExpr := fmt.Sprintf(
+			"arrayMap((tall, tok, vok) -> if(tall IS NULL OR tok IS NULL OR tall != tok, NULL, vok), timeSeriesLastToGrid%s(fromUnixTimestamp64Milli(unix_milli), toFloat64(unix_milli)), timeSeriesLastToGridIf%s(fromUnixTimestamp64Milli(unix_milli), toFloat64(unix_milli), bitAnd(flags, 1) = 0), timeSeriesLastToGridIf%s(fromUnixTimestamp64Milli(unix_milli), value, bitAnd(flags, 1) = 0))",
+			gridParams, gridParams, gridParams,
+		)
+		inner, innerArgs, err = joinedInner(gridExpr, false)
+	case unitOverTime:
+		if unit.overFn == "last" {
+			// last_over_time == last non-stale sample in the window: the
+			// stale rows are already excluded in WHERE.
+			gridExpr := fmt.Sprintf(
+				"timeSeriesLastToGrid(fromUnixTimestamp64Milli(%d), fromUnixTimestamp64Milli(%d), %d, %d)(fromUnixTimestamp64Milli(unix_milli), value)",
+				selStart, selEnd, stepSec, windowSec,
+			)
+			inner, innerArgs, err = joinedInner(gridExpr, true)
+			break
+		}
+		inner, innerArgs, err = windowedInner(unit, samplesConditions, seriesSub, inlineFingerprints == nil, adjustedTsStart, dataEnd, tsTable, selStart, selEnd, stepMs, windowMs)
+	default: // unitRange
+		gridExpr := fmt.Sprintf(
+			"%s(fromUnixTimestamp64Milli(%d), fromUnixTimestamp64Milli(%d), %d, %d)(fromUnixTimestamp64Milli(unix_milli), value)",
+			gridFunction[unit.fn], selStart, selEnd, stepSec, windowSec,
+		)
+		if unit.fn == fnIncrease {
+			// increase == rate * range-seconds, exactly: extrapolatedRate
+			// divides by the range only when isRate.
+			gridExpr = fmt.Sprintf("arrayMap(x -> x * %d, %s)", windowSec, gridExpr)
+		}
+		inner, innerArgs, err = joinedInner(gridExpr, true)
+	}
+	if err != nil {
+		return "", nil, err
+	}
+
+	spatial := "maxForEach(grid)"
+	switch {
+	case !unit.hasAgg:
+		// Per-series output: one row per (labels-minus-__name__) group.
+		// Distinct fingerprints can collapse onto the same projected label
+		// set only via a regex __name__ selector over metrics with identical
+		// other labels; maxForEach is a deterministic NULL-skipping merge and
+		// the identity for the overwhelmingly common one-fingerprint group.
+	case unit.aggOp.String() == "count":
+		// count over an all-absent index is an absent point, not 0.
+		spatial = "arrayMap(c -> if(c = 0, NULL, toFloat64(c)), countForEach(grid))"
+	default:
+		spatial = fmt.Sprintf("%s(grid)", aggForEach[unit.aggOp.String()])
+	}
+
+	query := fmt.Sprintf("SELECT gkey, %s AS grid FROM (%s) GROUP BY gkey %s", spatial, inner, gridFunctionsSetting)
+	return query, innerArgs, nil
+}
+
+// windowedInner builds the avg/min/max/sum/count _over_time form: each
+// sample fans out to every grid index k whose window (t_k - range, t_k]
+// contains it (ARRAY JOIN), aggregates per (fingerprint, k) — shard-side
+// partials over the distributed table — then assembles the positional grid
+// and joins the group key at the initiator over the reduced rows. The
+// group-key subquery reads the distributed series table here because it does
+// not ride inside a shard-rewritten query.
+//
+// This is the one form whose samples query has no series join, so an
+// over-the-limit fingerprint set (semiJoin) must fall back to the
+// shard-local semi-join: without it the fan-out would expand every series of
+// the metric and discard the unmatched ones only at the group-key join.
+func windowedInner(unit *coreUnit, samplesConditions func(*sqlbuilder.SelectBuilder, bool), seriesSub func(string) (string, []any, error), semiJoin bool, adjustedTsStart, dataEnd int64, tsTable string, selStart, selEnd, stepMs, windowMs int64) (string, []any, error) {
+	aggExpr := map[string]string{
+		"avg":   "avg(value)",
+		"min":   "min(value)",
+		"max":   "max(value)",
+		"sum":   "sum(value)",
+		"count": "toFloat64(count(value))",
+	}[unit.overFn]
+	effStepMs := stepMs
+	if effStepMs == 0 {
+		effStepMs = 1000
+	}
+	lastIdx := (selEnd - selStart) / effStepMs
+
+	perWindow := sqlbuilder.NewSelectBuilder()
+	perWindow.Select("fingerprint", "k", aggExpr+" AS v")
+	perWindow.From(fmt.Sprintf(
+		"%s.%s AS points ARRAY JOIN range(toUInt64(greatest(0, intDiv(unix_milli - %d + %d - 1, %d))), toUInt64(least(%d, intDiv(unix_milli + %d - 1 - %d, %d)) + 1)) AS k",
+		databaseName, distributedSamplesV4,
+		selStart, effStepMs, effStepMs,
+		lastIdx, windowMs, selStart, effStepMs,
+	))
+	samplesConditions(perWindow, true)
+	if semiJoin {
+		sub := sqlbuilder.NewSelectBuilder()
+		sub.Select("fingerprint")
+		sub.From(fmt.Sprintf("%s.%s", databaseName, localTimeSeriesTable(tsTable)))
+		if err := applySeriesConditions(sub, adjustedTsStart, dataEnd, unit.matchers); err != nil {
+			return "", nil, err
+		}
+		perWindow.Where(perWindow.In("points.fingerprint", sub))
+	}
+	perWindow.GroupBy("fingerprint", "k")
+	perWindowSQL, perWindowArgs := perWindow.BuildWithFlavor(sqlbuilder.ClickHouse)
+
+	grids := fmt.Sprintf(
+		"SELECT fingerprint, arrayMap(i -> if(indexOf(ks, i) = 0, NULL, vs[indexOf(ks, i)]), range(toUInt64(%d))) AS grid FROM (SELECT fingerprint, groupArray(k) AS ks, groupArray(v) AS vs FROM (%s) GROUP BY fingerprint)",
+		lastIdx+1, perWindowSQL,
+	)
+
+	seriesSQL, seriesArgs, err := seriesSub(tsTable)
+	if err != nil {
+		return "", nil, err
+	}
+	inner := fmt.Sprintf(
+		"SELECT series.gkey AS gkey, points.grid AS grid FROM (%s) AS points INNER JOIN (%s) AS series ON points.fingerprint = series.fingerprint",
+		grids, seriesSQL,
+	)
+	return inner, append(perWindowArgs, seriesArgs...), nil
+}
+
+// groupKeyExpr renders the canonical group key for a unit: the sorted
+// [key, value] pairs of the projected labels, JSON-encoded.
+//   - by (a, b): keep only the listed labels (absent labels stay absent,
+//     matching PromQL's by() over missing labels);
+//   - without (a, b): keep everything except the listed labels and __name__;
+//   - no aggregation: keep everything including __name__ — even when the
+//     unit drops the name from its OUTPUT, the key must keep it so distinct
+//     metrics never merge in SQL; executeUnit strips the name afterwards and
+//     turns a post-strip collision into the engine's duplicate-labelset
+//     error instead of a silently invented merge.
+func groupKeyExpr(unit *coreUnit) string {
+	// An empty label value means "label absent" in Prometheus; the stored
+	// labels JSON can carry empty attribute values, which must not become
+	// output labels or group keys.
+	pairs := "arraySort(JSONExtractKeysAndValues(labels, 'String'))"
+	if !unit.hasAgg {
+		return fmt.Sprintf("toJSONString(arrayFilter(p -> p.2 != '', %s))", pairs)
+	}
+	if unit.by {
+		if len(unit.grouping) == 0 {
+			return "'[]'"
+		}
+		return fmt.Sprintf("toJSONString(arrayFilter(p -> p.2 != '' AND p.1 IN (%s), %s))", quotedList(unit.grouping), pairs)
+	}
+	excluded := append([]string{metricNameLabel}, unit.grouping...)
+	return fmt.Sprintf("toJSONString(arrayFilter(p -> p.2 != '' AND p.1 NOT IN (%s), %s))", quotedList(excluded), pairs)
+}
+
+func quotedList(items []string) string {
+	quoted := make([]string, len(items))
+	for i, s := range items {
+		quoted[i] = "'" + strings.ReplaceAll(s, "'", "\\'") + "'"
+	}
+	return strings.Join(quoted, ", ")
+}

pkg/prometheus/clickhouseprometheusv2/transpiler_test.go

@@ -0,0 +1,517 @@
+package clickhouseprometheusv2
+
+import (
+	"context"
+	"sync/atomic"
+	"testing"
+	"time"
+
+	cmock "github.com/SigNoz/clickhouse-go-mock"
+	"github.com/SigNoz/signoz/pkg/errors"
+	"github.com/SigNoz/signoz/pkg/prometheus"
+	"github.com/prometheus/prometheus/model/labels"
+	"github.com/prometheus/prometheus/promql/parser"
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+)
+
+func parse(t *testing.T, q string) parser.Expr {
+	t.Helper()
+	expr, err := parser.NewParser(parser.Options{}).ParseExpr(q)
+	require.NoError(t, err)
+	return expr
+}
+
+func TestClassifyFullShapes(t *testing.T) {
+	tests := []struct {
+		name  string
+		query string
+		check func(t *testing.T, u *coreUnit)
+	}{
+		{
+			name:  "sum by rate",
+			query: `sum by (pod) (rate(http_requests_total{job="api"}[5m]))`,
+			check: func(t *testing.T, u *coreUnit) {
+				assert.Equal(t, fnRate, u.fn)
+				assert.Equal(t, int64(300_000), u.rangeMs)
+				assert.True(t, u.hasAgg)
+				assert.True(t, u.by)
+				assert.Equal(t, []string{"pod"}, u.grouping)
+			},
+		},
+		{
+			name:  "bare increase with offset",
+			query: `increase(errors_total[10m] offset 30m)`,
+			check: func(t *testing.T, u *coreUnit) {
+				assert.Equal(t, fnIncrease, u.fn)
+				assert.Equal(t, int64(1_800_000), u.offsetMs)
+				assert.False(t, u.hasAgg)
+			},
+		},
+		{
+			name:  "avg without over delta",
+			query: `avg without (instance) (delta(gauge_metric[15m]))`,
+			check: func(t *testing.T, u *coreUnit) {
+				assert.Equal(t, fnDelta, u.fn)
+				assert.True(t, u.hasAgg)
+				assert.False(t, u.by)
+			},
+		},
+		{
+			name:  "scalar pipeline with comparison",
+			query: `sum(rate(x[5m])) * 100 > 5`,
+			check: func(t *testing.T, u *coreUnit) {
+				require.Len(t, u.ops, 2)
+				assert.Equal(t, parser.ItemType(parser.MUL), u.ops[0].op)
+				assert.Equal(t, 100.0, u.ops[0].scalar)
+				assert.Equal(t, parser.ItemType(parser.GTR), u.ops[1].op)
+			},
+		},
+		{
+			name:  "scalar on left with unary minus",
+			query: `-1 * sum(rate(x[5m]))`,
+			check: func(t *testing.T, u *coreUnit) {
+				require.Len(t, u.ops, 1)
+				assert.True(t, u.ops[0].scalarOnLeft)
+				assert.Equal(t, -1.0, u.ops[0].scalar)
+			},
+		},
+		{
+			name:  "bool comparison",
+			query: `sum(rate(x[5m])) >= bool 0.5`,
+			check: func(t *testing.T, u *coreUnit) {
+				require.Len(t, u.ops, 1)
+				assert.True(t, u.ops[0].returnBool)
+			},
+		},
+		{
+			name:  "irate utf8 name",
+			query: `sum by ("k8s.pod.name") (irate({"k8s.container.cpu.time"}[2m]))`,
+			check: func(t *testing.T, u *coreUnit) {
+				assert.Equal(t, fnIRate, u.fn)
+				assert.Equal(t, []string{"k8s.pod.name"}, u.grouping)
+			},
+		},
+		{
+			name:  "bare instant selector keeps name",
+			query: `up{job="api"}`,
+			check: func(t *testing.T, u *coreUnit) {
+				assert.Equal(t, unitInstant, u.kind)
+				assert.True(t, u.keepsName())
+			},
+		},
+		{
+			name:  "gauge aggregation",
+			query: `sum by (pod) (container_memory offset 5m)`,
+			check: func(t *testing.T, u *coreUnit) {
+				assert.Equal(t, unitInstant, u.kind)
+				assert.Equal(t, int64(300_000), u.offsetMs)
+				assert.True(t, u.hasAgg)
+				assert.False(t, u.keepsName())
+			},
+		},
+		{
+			name:  "gauge comparison keeps name",
+			query: `container_memory > 100`,
+			check: func(t *testing.T, u *coreUnit) {
+				assert.Equal(t, unitInstant, u.kind)
+				assert.True(t, u.keepsName())
+			},
+		},
+		{
+			name:  "gauge arithmetic drops name",
+			query: `container_memory / 1024`,
+			check: func(t *testing.T, u *coreUnit) {
+				assert.Equal(t, unitInstant, u.kind)
+				assert.False(t, u.keepsName())
+			},
+		},
+		{
+			name:  "avg_over_time",
+			query: `max by (node) (avg_over_time(load1[10m]))`,
+			check: func(t *testing.T, u *coreUnit) {
+				assert.Equal(t, unitOverTime, u.kind)
+				assert.Equal(t, "avg", u.overFn)
+				assert.Equal(t, int64(600_000), u.rangeMs)
+			},
+		},
+		{
+			name:  "last_over_time keeps name",
+			query: `last_over_time(load1[10m])`,
+			check: func(t *testing.T, u *coreUnit) {
+				assert.Equal(t, unitOverTime, u.kind)
+				assert.Equal(t, "last", u.overFn)
+				assert.True(t, u.keepsName())
+			},
+		},
+	}
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			plan, ok := classify(parse(t, tt.query), testGrid(60_000))
+			require.True(t, ok, "expected transpilable")
+			require.True(t, plan.full, "expected full compilation")
+			require.Len(t, plan.units, 1)
+			tt.check(t, &plan.units[0].core)
+		})
+	}
+}
+
+func TestClassifyFallbackShapes(t *testing.T) {
+	queries := []struct {
+		name  string
+		query string
+		step  int64
+	}{
+		{"default-resolution subquery", `max_over_time(rate(x[5m])[30m:])`, 60_000},
+		{"at modifier", `sum(rate(x[5m] @ 1609746000))`, 60_000},
+		{"at modifier on gauge", `sum(container_memory @ 1609746000)`, 60_000},
+		{"sub-second step", `sum(rate(x[5m]))`, 500},
+		{"sub-second range", `sum(rate(x[1500ms]))`, 60_000},
+		{"by __name__ full", `sum by (__name__) (rate({__name__=~"a|b"}[5m]))`, 60_000},
+		{"quantile_over_time unsupported", `quantile_over_time(0.9, load1[10m])`, 60_000},
+	}
+	for _, tt := range queries {
+		t.Run(tt.name, func(t *testing.T) {
+			_, ok := classify(parse(t, tt.query), testGrid(tt.step))
+			assert.False(t, ok, "expected fallback for %s", tt.query)
+		})
+	}
+}
+
+func TestClassifyHybridShapes(t *testing.T) {
+	tests := []struct {
+		name          string
+		query         string
+		wantUnits     int
+		wantRewritten string
+	}{
+		{
+			name:          "histogram quantile",
+			query:         `histogram_quantile(0.95, sum by (le) (rate(http_bucket[5m])))`,
+			wantUnits:     1,
+			wantRewritten: `histogram_quantile(0.95, __signoz_transpiled_0__)`,
+		},
+		{
+			name:          "topk over compiled",
+			query:         `topk(5, sum by (pod) (rate(x[5m])))`,
+			wantUnits:     1,
+			wantRewritten: `topk(5, __signoz_transpiled_0__)`,
+		},
+		{
+			name:          "ratio of compiled units",
+			query:         `sum(rate(a[5m])) / sum(rate(b[5m]))`,
+			wantUnits:     2,
+			wantRewritten: `__signoz_transpiled_0__ / __signoz_transpiled_1__`,
+		},
+		{
+			name:          "or vector zero",
+			query:         `sum(rate(a[5m])) or vector(0)`,
+			wantUnits:     1,
+			wantRewritten: `__signoz_transpiled_0__ or vector(0)`,
+		},
+		{
+			name:          "quantile agg over compiled rate",
+			query:         `quantile(0.9, rate(x[5m]))`,
+			wantUnits:     1,
+			wantRewritten: `quantile(0.9, __signoz_transpiled_0__)`,
+		},
+		{
+			name:          "non-literal scalar side stays engine-side",
+			query:         `sum(rate(x[5m])) * scalar(y)`,
+			wantUnits:     1,
+			wantRewritten: `__signoz_transpiled_0__ * scalar(y)`,
+		},
+		{
+			name:          "compiled mixed with raw selector",
+			query:         `sum by (pod) (rate(a[5m])) / on (pod) group_left () b`,
+			wantUnits:     1,
+			wantRewritten: `__signoz_transpiled_0__ / on (pod) group_left () b`,
+		},
+	}
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			plan, ok := classify(parse(t, tt.query), testGrid(60_000))
+			require.True(t, ok)
+			assert.False(t, plan.full)
+			assert.Len(t, plan.units, tt.wantUnits)
+			assert.Equal(t, tt.wantRewritten, plan.rewritten)
+		})
+	}
+}
+
+func TestClassifyHybridGuards(t *testing.T) {
+	t.Run("no substitution under on(__name__)", func(t *testing.T) {
+		plan, ok := classify(parse(t, `sum(rate(a[5m])) * on (__name__) b`), testGrid(60_000))
+		_ = plan
+		assert.False(t, ok, "matching on __name__ must not see synthetic names")
+	})
+	t.Run("no substitution inside @-pinned subquery", func(t *testing.T) {
+		_, ok := classify(parse(t, `max_over_time(rate(x[5m])[30m:1m] @ 1609746000)`), testGrid(60_000))
+		assert.False(t, ok)
+	})
+}
+
+// The alert-smoothing idiom: units inside a fixed-resolution subquery
+// evaluate on the subquery grid — epoch-aligned multiples of the resolution,
+// starting strictly after (outer start - range), exactly as the engine
+// derives it.
+func TestClassifySubqueryUnits(t *testing.T) {
+	grid := gridContext{startMs: 1_700_000_030_000, endMs: 1_700_007_200_000, stepMs: 60_000}
+
+	plan, ok := classify(parse(t, `min_over_time((sum by (ns) (increase(x[5m])))[10m:5m]) > 0`), grid)
+	require.True(t, ok)
+	require.False(t, plan.full)
+	require.Len(t, plan.units, 1)
+	assert.Equal(t, `min_over_time(__signoz_transpiled_0__[10m:5m]) > 0`, plan.rewritten)
+
+	unit := plan.units[0]
+	// lower bound = outer start - range = 1_699_999_430_000; first multiple
+	// of 300_000 strictly greater is 1_699_999_500_000.
+	assert.Equal(t, int64(1_699_999_500_000), unit.grid.startMs)
+	assert.Equal(t, grid.endMs, unit.grid.endMs)
+	assert.Equal(t, int64(300_000), unit.grid.stepMs)
+	assert.Equal(t, fnIncrease, unit.core.fn)
+
+	t.Run("subquery offset shifts the grid", func(t *testing.T) {
+		plan, ok := classify(parse(t, `max_over_time((sum(rate(x[5m])))[10m:5m] offset 30m)`), grid)
+		require.True(t, ok)
+		require.Len(t, plan.units, 1)
+		// lower = start - offset - range = 1_699_997_630_000 -> first
+		// multiple of 300_000 above = 1_699_997_700_000; end shifts too.
+		assert.Equal(t, int64(1_699_997_700_000), plan.units[0].grid.startMs)
+		assert.Equal(t, grid.endMs-1_800_000, plan.units[0].grid.endMs)
+	})
+
+	t.Run("mollusk ratio-inside-subquery idiom", func(t *testing.T) {
+		q := `min_over_time(((sum by (a) (rate(m1[5m]))) / (avg by (a) (m2)))[5m:1m])`
+		plan, ok := classify(parse(t, q), grid)
+		require.True(t, ok)
+		// Both sides compile on the subquery grid: the rate side and the
+		// gauge aggregation side; the engine joins them and smooths.
+		require.Len(t, plan.units, 2)
+		assert.Equal(t, int64(60_000), plan.units[0].grid.stepMs)
+		assert.Equal(t, unitInstant, plan.units[1].core.kind)
+		assert.Contains(t, plan.rewritten, `__signoz_transpiled_0__ / __signoz_transpiled_1__`)
+	})
+}
+
+func TestBuildUnitSQL(t *testing.T) {
+	unit := &coreUnit{
+		fn:       fnRate,
+		rangeMs:  300_000,
+		hasAgg:   true,
+		aggOp:    parser.SUM,
+		by:       true,
+		grouping: []string{"pod"},
+		matchers: []*labels.Matcher{mustMatcher(t, labels.MatchEqual, "__name__", "http_requests_total")},
+	}
+	sql, args, err := buildUnitSQL(unit, []string{"http_requests_total"}, []uint64{7, 42}, 1_699_999_700_000, 1_700_003_600_000, 1_700_000_000_000, 1_700_003_600_000, 60_000, 300_000)
+	require.NoError(t, err)
+
+	assert.Contains(t, sql, "timeSeriesRateToGrid(fromUnixTimestamp64Milli(1700000000000), fromUnixTimestamp64Milli(1700003600000), 60, 300)(fromUnixTimestamp64Milli(unix_milli), value)")
+	assert.Contains(t, sql, "unix_milli > ? AND unix_milli <= ?")
+	assert.Contains(t, sql, "bitAnd(flags, 1) = 0")
+	assert.Contains(t, sql, "sumForEach(grid)")
+	// The group-key join rides inside the shard query: distributed samples
+	// at the top level, the local series table in the join subquery, the
+	// grid aggregation grouped per (fingerprint, gkey) shard-side.
+	assert.Contains(t, sql, "FROM signoz_metrics.distributed_samples_v4 AS points INNER JOIN (SELECT fingerprint,")
+	assert.Contains(t, sql, "FROM signoz_metrics.time_series_v4 WHERE")
+	assert.Contains(t, sql, "GROUP BY points.fingerprint, series.gkey")
+	assert.Contains(t, sql, "points.fingerprint IN (7, 42)")
+	assert.Contains(t, sql, `toJSONString(arrayFilter(p -> p.2 != '' AND p.1 IN ('pod'),`)
+	assert.Contains(t, sql, "SETTINGS allow_experimental_ts_to_grid_aggregate_function = 1")
+	// Args follow placeholder order: the joined series subquery renders
+	// before the samples WHERE.
+	assert.Equal(t, []any{"http_requests_total", int64(1_699_999_200_000), int64(1_700_003_600_000), "http_requests_total", int64(1_699_999_700_000), int64(1_700_003_600_000)}, args)
+}
+
+func TestBuildUnitSQLIncreaseAndOffset(t *testing.T) {
+	unit := &coreUnit{
+		fn:       fnIncrease,
+		rangeMs:  600_000,
+		offsetMs: 1_800_000,
+		matchers: []*labels.Matcher{mustMatcher(t, labels.MatchEqual, "__name__", "errors_total")},
+	}
+	sql, _, err := buildUnitSQL(unit, nil, []uint64{7}, 1_699_997_600_000, 1_700_001_800_000, 1_700_000_000_000, 1_700_003_600_000, 60_000, 300_000)
+	require.NoError(t, err)
+
+	// Grid and window shift by the offset; increase multiplies rate by the
+	// range in seconds.
+	assert.Contains(t, sql, "fromUnixTimestamp64Milli(1699998200000), fromUnixTimestamp64Milli(1700001800000)")
+	assert.Contains(t, sql, "arrayMap(x -> x * 600, timeSeriesRateToGrid")
+	assert.Contains(t, sql, "maxForEach(grid)")
+}
+
+func TestBuildUnitSQLOverLimitJoinOnly(t *testing.T) {
+	// Past the inline limit no fingerprint filter is rendered: the series
+	// join restricts to the matched fingerprints on its own.
+	unit := &coreUnit{
+		fn:       fnRate,
+		rangeMs:  300_000,
+		hasAgg:   true,
+		aggOp:    parser.SUM,
+		by:       true,
+		matchers: []*labels.Matcher{mustMatcher(t, labels.MatchEqual, "__name__", "http_requests_total")},
+	}
+	sql, _, err := buildUnitSQL(unit, []string{"http_requests_total"}, nil, 1_699_999_700_000, 1_700_003_600_000, 1_700_000_000_000, 1_700_003_600_000, 60_000, 300_000)
+	require.NoError(t, err)
+
+	assert.NotContains(t, sql, "points.fingerprint IN")
+	assert.Contains(t, sql, "INNER JOIN (SELECT fingerprint,")
+	assert.Contains(t, sql, "FROM signoz_metrics.time_series_v4 WHERE")
+}
+
+func TestBuildUnitSQLOverLimitWindowedSemiJoin(t *testing.T) {
+	// The windowed *_over_time fan-out has no series join, so the over-limit
+	// regime falls back to the shard-local semi-join instead of expanding
+	// every series of the metric.
+	unit := &coreUnit{
+		kind:     unitOverTime,
+		overFn:   "avg",
+		rangeMs:  600_000,
+		matchers: []*labels.Matcher{mustMatcher(t, labels.MatchEqual, "__name__", "node_load1")},
+	}
+	sql, _, err := buildUnitSQL(unit, []string{"node_load1"}, nil, 1_699_999_400_000, 1_700_003_600_000, 1_700_000_000_000, 1_700_003_600_000, 60_000, 300_000)
+	require.NoError(t, err)
+
+	assert.Contains(t, sql, "points.fingerprint IN (SELECT fingerprint FROM signoz_metrics.time_series_v4 WHERE ")
+	assert.Contains(t, sql, "ARRAY JOIN range(")
+}
+
+func TestApplyScalarOps(t *testing.T) {
+	f := func(v float64) *float64 { return &v }
+
+	t.Run("arithmetic chain", func(t *testing.T) {
+		values := []*float64{f(2), nil, f(4)}
+		applyScalarOps([]scalarOp{{op: parser.MUL, scalar: 100}, {op: parser.ADD, scalar: 1}}, values)
+		require.NotNil(t, values[0])
+		assert.Equal(t, 201.0, *values[0])
+		assert.Nil(t, values[1])
+		assert.Equal(t, 401.0, *values[2])
+	})
+
+	t.Run("comparison filters points", func(t *testing.T) {
+		values := []*float64{f(1), f(10)}
+		applyScalarOps([]scalarOp{{op: parser.GTR, scalar: 5}}, values)
+		assert.Nil(t, values[0])
+		require.NotNil(t, values[1])
+		assert.Equal(t, 10.0, *values[1], "filter comparisons keep the original value")
+	})
+
+	t.Run("bool comparison emits 0/1", func(t *testing.T) {
+		values := []*float64{f(1), f(10)}
+		applyScalarOps([]scalarOp{{op: parser.GTR, scalar: 5, returnBool: true}}, values)
+		assert.Equal(t, 0.0, *values[0])
+		assert.Equal(t, 1.0, *values[1])
+	})
+
+	t.Run("scalar on left division", func(t *testing.T) {
+		values := []*float64{f(4)}
+		applyScalarOps([]scalarOp{{op: parser.DIV, scalar: 100, scalarOnLeft: true}}, values)
+		assert.Equal(t, 25.0, *values[0])
+	})
+}
+
+func TestLabelsFromGroupKey(t *testing.T) {
+	lset, err := labelsFromGroupKey(`[["pod","api-0"],["ns","prod"]]`)
+	require.NoError(t, err)
+	assert.Equal(t, "api-0", lset.Get("pod"))
+	assert.Equal(t, "prod", lset.Get("ns"))
+
+	empty, err := labelsFromGroupKey(`[]`)
+	require.NoError(t, err)
+	assert.True(t, empty.IsEmpty())
+}
+
+// testGrid is a 2h query grid ending on a round timestamp.
+func testGrid(stepMs int64) gridContext {
+	return gridContext{startMs: 1_700_000_000_000, endMs: 1_700_007_200_000, stepMs: stepMs}
+}
+
+// A bool comparison returns 0/1, not the sample, so the engine drops
+// __name__; keeping it would change downstream vector matching.
+func TestKeepsName_BoolComparisonDropsName(t *testing.T) {
+	plan, ok := classify(parse(t, `up > bool 0`), testGrid(60_000))
+	require.True(t, ok)
+	assert.False(t, plan.units[0].core.keepsName())
+
+	plan, ok = classify(parse(t, `up > 0`), testGrid(60_000))
+	require.True(t, ok)
+	assert.True(t, plan.units[0].core.keepsName())
+}
+
+// timeSeriesLastToGrid widens its window to max(window, step) — probed on
+// 25.12 — so Last-style units at window < step must fall back or they would
+// resurrect samples the engine's lookback already dropped.
+func TestTryExecuteRange_LastStyleWindowBelowStepFallsBack(t *testing.T) {
+	c, _ := newTestClient(t, prometheus.ClickhouseV2Config{})
+	e := &executor{client: c, parser: prometheus.NewParser()}
+
+	start := time.UnixMilli(1_700_000_000_000)
+	end := time.UnixMilli(1_700_003_600_000)
+
+	_, ok, err := e.TryExecuteRange(context.Background(), `sum by (pod) (up)`, start, end, time.Hour)
+	require.NoError(t, err)
+	assert.False(t, ok, "instant selection at step > lookback must not transpile")
+
+	_, ok, err = e.TryExecuteRange(context.Background(), `last_over_time(up[10m])`, start, end, time.Hour)
+	require.NoError(t, err)
+	assert.False(t, ok, "last_over_time at range < step must not transpile")
+}
+
+// Transpiled results never pass the engine's sample limiter, so the grid
+// cells (series x grid width) must be budgeted before the arrays exist —
+// otherwise a wide query rebuilds the OOM this provider exists to prevent.
+func TestExecuteUnit_GridCellBudget(t *testing.T) {
+	c, store := newTestClient(t, prometheus.ClickhouseV2Config{MaxFetchedSamples: 100})
+	e := &executor{client: c, parser: prometheus.NewParser()}
+
+	store.Mock().ExpectQuery("SELECT fingerprint, any\\(labels\\)").WithArgs("up", int64(1_699_999_200_000), int64(1_700_003_600_000)).WillReturnRows(cmock.NewRows(seriesCols, [][]any{
+		{uint64(1), `{"__name__":"up","instance":"a"}`},
+		{uint64(2), `{"__name__":"up","instance":"b"}`},
+	}))
+
+	plan, ok := classify(parse(t, `sum(rate(up[5m]))`), gridContext{startMs: 1_700_000_000_000, endMs: 1_700_003_600_000, stepMs: 60_000})
+	require.True(t, ok)
+
+	var cells atomic.Int64
+	// 2 series x 61 grid points = 122 cells > 100.
+	_, err := e.executeUnit(context.Background(), &plan.units[0].core, plan.units[0].grid, &cells)
+	require.Error(t, err)
+	assert.True(t, errors.Ast(err, errors.TypeInvalidInput), "budget refusal must be typed invalid input, got %v", err)
+}
+
+// Two metrics collapsing onto one labelset after the name drop is the
+// engine's duplicate-labelset error; silently merging them would invent a
+// series no engine would produce.
+func TestExecuteUnit_NameCollisionErrors(t *testing.T) {
+	c, store := newTestClient(t, prometheus.ClickhouseV2Config{})
+	e := &executor{client: c, parser: prometheus.NewParser()}
+
+	store.Mock().ExpectQuery("SELECT fingerprint, any\\(labels\\)").WithArgs("^(?:a|b)$", int64(1_699_999_200_000), int64(1_700_003_600_000)).WillReturnRows(cmock.NewRows(seriesCols, [][]any{
+		{uint64(1), `{"__name__":"a","job":"x"}`},
+		{uint64(2), `{"__name__":"b","job":"x"}`},
+	}))
+	store.Mock().ExpectQuery("SELECT gkey").
+		WithArgs("^(?:a|b)$", int64(1_699_999_200_000), int64(1_700_003_600_000), "a", "b", int64(1_699_999_700_000), int64(1_700_003_600_000)).
+		WillReturnRows(cmock.NewRows(gkeyCols, [][]any{
+			{`[["__name__","a"],["job","x"]]`, []*float64{f64(1)}},
+			{`[["__name__","b"],["job","x"]]`, []*float64{f64(2)}},
+		}))
+
+	plan, ok := classify(parse(t, `rate({__name__=~"a|b"}[5m])`), gridContext{startMs: 1_700_000_000_000, endMs: 1_700_003_600_000, stepMs: 60_000})
+	require.True(t, ok)
+
+	var cells atomic.Int64
+	_, err := e.executeUnit(context.Background(), &plan.units[0].core, plan.units[0].grid, &cells)
+	require.Error(t, err)
+	assert.Contains(t, err.Error(), "vector cannot contain metrics with the same labelset")
+}
+
+var gkeyCols = []cmock.ColumnType{
+	{Name: "gkey", Type: "String"},
+	{Name: "grid", Type: "Array(Nullable(Float64))"},
+}
+
+func f64(v float64) *float64 { return &v }