[hypequery](https://hypequery.com/) › [Recipes](https://hypequery.com/recipes) › Cursor pagination in ClickHouse: keyset vs LIMIT/OFFSET, measured Index for agents: https://hypequery.com/llms.txt # Cursor pagination in ClickHouse: keyset vs LIMIT/OFFSET, measured (https://hypequery.com/recipes/cursor-pagination) Page a tenant's events newest first with a keyset cursor on (occurred_at, event_id) and see from EXPLAIN and query_log why it beats OFFSET on deep pages. ## Run it ```bash npx @hypequery/cli create cursor-pagination -y cd cursor-pagination pnpm install pnpm dev ``` - Packages: @hypequery/clickhouse (tested from hypequery 2.14.0) - Stack: node, clickhouse - Source: https://github.com/hypequery/hypequery/tree/main/recipes/cursor-pagination - Test: `pnpm test` - Verify: `node dist/recipe.js --json` | Env var | Required | Description | | --- | --- | --- | | CLICKHOUSE_URL | no | ClickHouse HTTP endpoint (default http://localhost:8123) | | CLICKHOUSE_USERNAME | no | ClickHouse user (default: default) | | CLICKHOUSE_PASSWORD | no | ClickHouse password (default: hypequery, the seed's password) | | CLICKHOUSE_DATABASE | no | Database holding the seed tables (default hypequery_examples) | For cursor pagination in ClickHouse, order by a unique key that follows the table's sort key, `ORDER BY occurred_at DESC, event_id DESC`, and fetch the next page with `WHERE occurred_at < :t OR (occurred_at = :t AND event_id < :id)`, where `:t, :id` is the last row of the previous page. On the seed, page 1,001 of one tenant read about 6,000 to 9,000 rows that way and about 105,000 rows with `OFFSET 50000`, the tenant's whole history. Write the condition as the expanded `OR`, not as the tuple `(occurred_at, event_id) < (:t, :id)`: on ClickHouse 25.8 the tuple form is not used by partition pruning or the primary key. Longer background: [How to paginate ClickHouse results in TypeScript](https://hypequery.com/blog/clickhouse-pagination-typescript). ## Demo ```text Tenant 47, newest first, 50 per page page 1 825213 2026-10-31 23:58:36 api_call /api/v1/datasets 777014 2026-10-31 23:44:38 page_view /dashboard ... 499194 2026-10-31 16:49:06 login /login <- cursor page 2 196645 2026-10-31 16:46:30 page_view /explore 635639 2026-10-31 16:44:31 api_call /api/v1/query ... 508131 2026-10-31 12:14:23 api_call /api/v1/users <- next cursor OFFSET vs keyset, 50 rows a page page offset granules after the primary key rows read (query_log) OFFSET keyset OFFSET keyset 1 0 106 106 9-12k 9k 21 1,000 106 105 27k 9-12k 201 10,000 106 96 89-104k 9-12k 1001 50,000 106 54 105k 6-9k ``` Granule counts come from `EXPLAIN indexes = 1` and are pinned by the test. Rows read come from `system.query_log` and move by a granule or two between runs with `max_threads` (8 cores here), so the test asserts their shape, not exact values. ## Install ```bash npx @hypequery/cli create cursor-pagination -y cd cursor-pagination docker compose -f seed/docker-compose.yml up -d --wait && (cd seed && pnpm install && pnpm seed) # or set CLICKHOUSE_* for your own ClickHouse pnpm install pnpm dev ``` `pnpm dev` prints the first two pages, then builds `cursor_pagination_events` (a compacted copy of `events`, see below), prints the cost table and drops the copy again. ## What it shows - **A keyset page with the typed query builder** (`src/pagination.ts`): `whereGroup` / `orWhereGroup` build the `OR` condition, and the cursor is an opaque base64url string of `{ occurredAt, eventId }`. - **The generated SQL is asserted** (`expected.sql`): page 2's statement, with the cursor bound as parameters. - **The first two pages are pinned** (`output.json`), and the test walks five pages both ways to prove keyset and OFFSET return the same rows. - **Ties are handled**: tenant 47 has three events at 2026-10-30 14:03:47; a cursor on `occurred_at` alone skips the rest of that second, the `(occurred_at, event_id)` cursor does not. The test checks both. - **Cost, measured** (`src/cost.ts`): granules from `EXPLAIN indexes = 1`, rows read from `system.query_log` by `queryId`. ## Key file ```ts export function keysetPageQuery(tenantId: number, pageSize: number, after: Cursor | null) { let query = db.table('events') .select(['event_id', 'occurred_at', 'event_type', 'user_id', 'path']) .where('tenant_id', 'eq', tenantId); if (after) { query = query.whereGroup((q) => { q.where('occurred_at', 'lt', after.occurredAt).orWhereGroup((tie) => { tie.where('occurred_at', 'eq', after.occurredAt).where('event_id', 'lt', after.eventId); }); }); } return query.orderBy('occurred_at', 'DESC').orderBy('event_id', 'DESC').limit(pageSize); } ``` ## Why keyset wins, and when it does not - **OFFSET reads what it skips.** `LIMIT 50 OFFSET 50000` has to produce 50,050 rows in order before it can drop 50,000. Already at `OFFSET 10000` the server read 89,000 to 104,000 rows, close to all of the tenant's 97,376. - **Keyset skips with the index.** The cursor condition is a range on the sort key `(tenant_id, occurred_at, event_id)`, so partition pruning drops months after the cursor, the primary key narrows the granules (54 of 977 on page 1,001), and the plan reads in reverse sort-key order (`ReadType: InReverseOrder` in `EXPLAIN actions = 1`), so it stops once it has 50 rows. - **The tuple trap.** `(occurred_at, event_id) < ('2026-07-09 18:05:48', 735428)` leaves 106 of 977 granules (only `tenant_id` is used, all 12 parts scanned); the expanded `OR` leaves 54 of 620 after pruning 3 parts. Same rows, different plan. - **The sort key decides.** Keyset is fast here because `tenant_id` comes first in `ORDER BY`. The same cursor across all tenants gets partition pruning but almost nothing from the primary key (583 of the remaining 620 granules kept), because `occurred_at` is not a prefix of it. Give such a feed its own table or projection ordered by `(occurred_at, event_id)`. - **What you give up.** No "jump to page 47" and no total page count for free. For a feed or an API that is usually fine; for a small admin table OFFSET is simpler and cheap. ### Why a copy for the measurements The seed's `events` table holds a million rows in about 60 small parts with 8,192-row granules, and background merges change that layout over time. One granule then covers a large share of a tenant's month, so skipping barely shows, and the numbers drift. `src/cost.ts` copies the same rows into `cursor_pagination_events` with the same sort key, 1,024-row granules and one part per month (`OPTIMIZE ... FINAL`), so the comparison is stable. The pages themselves are read from `events`. ## Architecture ```text client ── GET /events?cursor= │ src/pagination.ts keysetPage(tenant, 50, cursor) │ WHERE tenant_id = ? AND (occurred_at < ? OR (occurred_at = ? AND event_id < ?)) │ ORDER BY occurred_at DESC, event_id DESC LIMIT 50 ▼ { rows, nextCursor } ──▶ output.json (pages 1 and 2, asserted in CI) ``` ## Adapt it - **Oldest first.** Flip both `DESC` to `ASC` and `lt` to `gt`. - **Another sort.** The cursor must be the full `ORDER BY` plus a unique tie-breaker, in the same direction, and ideally a prefix-compatible slice of the table's sort key. - **Filters.** Add `where` clauses freely; they do not change the cursor. Changing the sort or filters invalidates old cursors, so put a version or the filter hash in the cursor if clients keep them. - **Serve it.** Return `{ rows, nextCursor }` from any handler; `@hypequery/serve` or the FastAPI example show the HTTP side. - **Aggregated lists.** Dataset queries already page grouped results with `limit` / `offset` and report `meta.pagination.hasMore`; those results are small, so OFFSET is fine there. ## Related examples - [multi-tenant-embedded-analytics](../../examples/multi-tenant-embedded-analytics): per-tenant APIs where a feed like this lives. - [fastapi-analytics-api](../../examples/fastapi-analytics-api): the HTTP side in Python. Tested on ClickHouse 25.8 with `@hypequery/clickhouse` 2.14. ## Key file: src/pagination.ts ```typescript import { db } from './client.js'; import type { Schema } from './schema.js'; export type EventsTable = keyof Schema; /** The last row of a page: where the next page starts. Opaque to the client. */ export interface Cursor { occurredAt: string; /** UInt64 arrives as a string from the JSON API; keep it one. */ eventId: string; } export const encodeCursor = (c: Cursor) => Buffer.from(JSON.stringify(c)).toString('base64url'); export const decodeCursor = (s: string): Cursor => JSON.parse(Buffer.from(s, 'base64url').toString('utf8')); const COLUMNS = ['event_id', 'occurred_at', 'event_type', 'user_id', 'path'] as const; /** * Keyset ("cursor") pagination, newest first: rows strictly before the * cursor in (occurred_at, event_id) order. The condition is spelled out as * `occurred_at < t OR (occurred_at = t AND event_id < id)` rather than the * tuple `(occurred_at, event_id) < (t, id)`: ClickHouse 25.8 uses the * expanded form for partition pruning and the primary key, and the tuple * form for neither. */ export function keysetPageQuery(tenantId: number, pageSize: number, after: Cursor | null, table: EventsTable = 'events') { let query = db .table(table) .select([...COLUMNS]) .where('tenant_id', 'eq', tenantId); if (after) { query = query.whereGroup((q) => { q.where('occurred_at', 'lt', after.occurredAt).orWhereGroup((tie) => { tie.where('occurred_at', 'eq', after.occurredAt).where('event_id', 'lt', after.eventId); }); }); } return query.orderBy('occurred_at', 'DESC').orderBy('event_id', 'DESC').limit(pageSize); } export async function keysetPage(tenantId: number, pageSize: number, after: string | null) { const rows = await keysetPageQuery(tenantId, pageSize, after ? decodeCursor(after) : null).execute(); const last = rows.at(-1); return { rows, // A short page is the last one, so no cursor; a full page may still be // the last, and the next call then returns zero rows. nextCursor: rows.length === pageSize && last ? encodeCursor({ occurredAt: String(last.occurred_at), eventId: String(last.event_id) }) : null, }; } /** LIMIT/OFFSET, for comparison: ClickHouse reads and discards every skipped row. */ export function offsetPageQuery(tenantId: number, pageSize: number, page: number, table: EventsTable = 'events') { return db .table(table) .select([...COLUMNS]) .where('tenant_id', 'eq', tenantId) .orderBy('occurred_at', 'DESC') .orderBy('event_id', 'DESC') .limit(pageSize) .offset(page * pageSize); } ``` ## Dataset: src/schema.ts ```typescript /** * The seed's `events` table as hypequery sees it (`npx hypequery generate` * writes this for you). Sort key: (tenant_id, occurred_at, event_id), so * "one tenant's events in time order" is a contiguous range of the index, * and (occurred_at, event_id) is a unique, ordered cursor within a tenant. * * `cursor_pagination_events` is the recipe's own compacted copy, used only to * measure what each pagination style reads (see src/cost.ts). */ type EventColumns = { event_id: 'UInt64'; tenant_id: 'UInt32'; user_id: 'UInt32'; event_type: 'LowCardinality(String)'; path: 'String'; status_code: 'UInt16'; latency_ms: 'UInt32'; device: 'LowCardinality(String)'; occurred_at: "DateTime('UTC')"; }; export type Schema = { events: EventColumns; cursor_pagination_events: EventColumns; }; ``` ## Generated SQL ```sql SELECT event_id, occurred_at, event_type, user_id, path FROM events WHERE tenant_id = ? AND (occurred_at < ? OR (occurred_at = ? AND event_id < ?)) ORDER BY occurred_at DESC, event_id DESC LIMIT 50 ``` ## Output on the seed dataset ```json { "page1": [ { "event_id": "825213", "occurred_at": "2026-10-31 23:58:36", "event_type": "api_call", "user_id": 1889, "path": "/api/v1/datasets" }, { "event_id": "777014", "occurred_at": "2026-10-31 23:44:38", "event_type": "page_view", "user_id": 1891, "path": "/dashboard" }, { "event_id": "935492", "occurred_at": "2026-10-31 23:33:03", "event_type": "login", "user_id": 2031, "path": "/login" }, { "event_id": "294756", "occurred_at": "2026-10-31 23:27:04", "event_type": "api_call", "user_id": 2058, "path": "/api/v1/query" }, { "event_id": "294478", "occurred_at": "2026-10-31 23:07:33", "event_type": "api_call", "user_id": 1909, "path": "/api/v1/export" }, { "event_id": "93102", "occurred_at": "2026-10-31 22:32:44", "event_type": "page_view", "user_id": 1903, "path": "/dashboard" }, { "event_id": "397022", "occurred_at": "2026-10-31 22:26:59", "event_type": "api_call", "user_id": 1889, "path": "/api/v1/events" }, { "event_id": "821664", "occurred_at": "2026-10-31 22:20:03", "event_type": "page_view", "user_id": 2059, "path": "/billing" }, { "event_id": "179828", "occurred_at": "2026-10-31 22:19:17", "event_type": "query_run", "user_id": 1905, "path": "/explore" }, { "event_id": "656327", "occurred_at": "2026-10-31 22:10:11", "event_type": "page_view", "user_id": 1963, "path": "/reports" }, { "event_id": "299529", "occurred_at": "2026-10-31 21:48:27", "event_type": "query_run", "user_id": 2003, "path": "/explore" }, { "event_id": "637913", "occurred_at": "2026-10-31 21:21:46", "event_type": "login", "user_id": 1968, "path": "/login" }, { "event_id": "780308", "occurred_at": "2026-10-31 21:13:57", "event_type": "page_view", "user_id": 1891, "path": "/alerts" }, { "event_id": "968988", "occurred_at": "2026-10-31 21:09:48", "event_type": "query_run", "user_id": 1972, "path": "/explore" }, { "event_id": "127334", "occurred_at": "2026-10-31 20:50:17", "event_type": "export", "user_id": 1972, "path": "/export" }, { "event_id": "555402", "occurred_at": "2026-10-31 20:49:16", "event_type": "login", "user_id": 1863, "path": "/login" }, { "event_id": "82528", "occurred_at": "2026-10-31 20:40:52", "event_type": "query_run", "user_id": 1905, "path": "/explore" }, { "event_id": "980412", "occurred_at": "2026-10-31 20:38:07", "event_type": "query_run", "user_id": 1979, "path": "/explore" }, { "event_id": "21641", "occurred_at": "2026-10-31 20:35:06", "event_type": "page_view", "user_id": 1984, "path": "/alerts" }, { "event_id": "489677", "occurred_at": "2026-10-31 20:30:10", "event_type": "page_view", "user_id": 1963, "path": "/dashboard" }, { "event_id": "774752", "occurred_at": "2026-10-31 20:26:42", "event_type": "page_view", "user_id": 2023, "path": "/settings" }, { "event_id": "908309", "occurred_at": "2026-10-31 20:19:41", "event_type": "api_call", "user_id": 1885, "path": "/api/v1/users" }, { "event_id": "669774", "occurred_at": "2026-10-31 20:02:40", "event_type": "query_run", "user_id": 1945, "path": "/explore" }, { "event_id": "829012", "occurred_at": "2026-10-31 20:01:07", "event_type": "page_view", "user_id": 1889, "path": "/billing" }, { "event_id": "716654", "occurred_at": "2026-10-31 20:00:53", "event_type": "query_run", "user_id": 1949, "path": "/explore" }, { "event_id": "738629", "occurred_at": "2026-10-31 20:00:49", "event_type": "page_view", "user_id": 1848, "path": "/explore" }, { "event_id": "901894", "occurred_at": "2026-10-31 19:31:55", "event_type": "page_view", "user_id": 1978, "path": "/reports" }, { "event_id": "704107", "occurred_at": "2026-10-31 19:19:05", "event_type": "api_call", "user_id": 1879, "path": "/api/v1/events" }, { "event_id": "966726", "occurred_at": "2026-10-31 19:05:31", "event_type": "login", "user_id": 2008, "path": "/login" }, { "event_id": "873359", "occurred_at": "2026-10-31 19:05:15", "event_type": "page_view", "user_id": 1889, "path": "/dashboard" }, { "event_id": "709106", "occurred_at": "2026-10-31 19:00:05", "event_type": "page_view", "user_id": 1841, "path": "/reports" }, { "event_id": "648913", "occurred_at": "2026-10-31 18:58:48", "event_type": "query_run", "user_id": 2059, "path": "/explore" }, { "event_id": "80483", "occurred_at": "2026-10-31 18:56:34", "event_type": "error", "user_id": 2058, "path": "/dashboard" }, { "event_id": "597547", "occurred_at": "2026-10-31 18:40:29", "event_type": "api_call", "user_id": 1979, "path": "/api/v1/users" }, { "event_id": "973335", "occurred_at": "2026-10-31 18:38:36", "event_type": "api_call", "user_id": 1889, "path": "/api/v1/users" }, { "event_id": "721142", "occurred_at": "2026-10-31 18:21:54", "event_type": "api_call", "user_id": 2008, "path": "/api/v1/export" }, { "event_id": "384713", "occurred_at": "2026-10-31 18:17:22", "event_type": "page_view", "user_id": 2028, "path": "/explore" }, { "event_id": "571022", "occurred_at": "2026-10-31 18:16:00", "event_type": "page_view", "user_id": 1933, "path": "/explore" }, { "event_id": "970250", "occurred_at": "2026-10-31 18:14:56", "event_type": "page_view", "user_id": 1979, "path": "/settings" }, { "event_id": "205863", "occurred_at": "2026-10-31 17:56:37", "event_type": "query_run", "user_id": 2023, "path": "/explore" }, { "event_id": "477621", "occurred_at": "2026-10-31 17:54:03", "event_type": "api_call", "user_id": 1972, "path": "/api/v1/export" }, { "event_id": "777899", "occurred_at": "2026-10-31 17:48:18", "event_type": "query_run", "user_id": 1935, "path": "/explore" }, { "event_id": "510795", "occurred_at": "2026-10-31 17:46:47", "event_type": "page_view", "user_id": 1859, "path": "/reports" }, { "event_id": "961248", "occurred_at": "2026-10-31 17:38:34", "event_type": "page_view", "user_id": 1933, "path": "/reports" }, { "event_id": "421069", "occurred_at": "2026-10-31 17:36:47", "event_type": "page_view", "user_id": 2033, "path": "/alerts" }, { "event_id": "239439", "occurred_at": "2026-10-31 17:25:42", "event_type": "api_call", "user_id": 1899, "path": "/api/v1/users" }, { "event_id": "907866", "occurred_at": "2026-10-31 17:22:43", "event_type": "page_view", "user_id": 1934, "path": "/alerts" }, { "event_id": "810745", "occurred_at": "2026-10-31 16:55:06", "event_type": "query_run", "user_id": 1909, "path": "/explore" }, { "event_id": "3037", "occurred_at": "2026-10-31 16:49:18", "event_type": "query_run", "user_id": 1841, "path": "/explore" }, { "event_id": "499194", "occurred_at": "2026-10-31 16:49:06", "event_type": "login", "user_id": 2059, "path": "/login" } ], "cursor": "eyJvY2N1cnJlZEF0IjoiMjAyNi0xMC0zMSAxNjo0OTowNiIsImV2ZW50SWQiOiI0OTkxOTQifQ", "page2": [ { "event_id": "196645", "occurred_at": "2026-10-31 16:46:30", "event_type": "page_view", "user_id": 1933, "path": "/explore" }, { "event_id": "635639", "occurred_at": "2026-10-31 16:44:31", "event_type": "api_call", "user_id": 2058, "path": "/api/v1/query" }, { "event_id": "541872", "occurred_at": "2026-10-31 16:44:07", "event_type": "query_run", "user_id": 2032, "path": "/explore" }, { "event_id": "105753", "occurred_at": "2026-10-31 16:42:46", "event_type": "api_call", "user_id": 1896, "path": "/api/v1/datasets" }, { "event_id": "259012", "occurred_at": "2026-10-31 16:40:25", "event_type": "api_call", "user_id": 1889, "path": "/api/v1/datasets" }, { "event_id": "432446", "occurred_at": "2026-10-31 16:37:14", "event_type": "login", "user_id": 2033, "path": "/login" }, { "event_id": "407452", "occurred_at": "2026-10-31 16:36:35", "event_type": "page_view", "user_id": 1879, "path": "/explore" }, { "event_id": "430566", "occurred_at": "2026-10-31 16:29:07", "event_type": "page_view", "user_id": 2008, "path": "/settings" }, { "event_id": "87885", "occurred_at": "2026-10-31 16:18:49", "event_type": "query_run", "user_id": 1936, "path": "/explore" }, { "event_id": "915802", "occurred_at": "2026-10-31 16:08:14", "event_type": "api_call", "user_id": 1891, "path": "/api/v1/query" }, { "event_id": "253611", "occurred_at": "2026-10-31 16:06:17", "event_type": "page_view", "user_id": 1842, "path": "/reports" }, { "event_id": "73579", "occurred_at": "2026-10-31 16:06:06", "event_type": "page_view", "user_id": 2003, "path": "/team" }, { "event_id": "794247", "occurred_at": "2026-10-31 15:56:09", "event_type": "query_run", "user_id": 1944, "path": "/explore" }, { "event_id": "528414", "occurred_at": "2026-10-31 15:48:16", "event_type": "api_call", "user_id": 2058, "path": "/api/v1/events" }, { "event_id": "736289", "occurred_at": "2026-10-31 15:46:39", "event_type": "page_view", "user_id": 1933, "path": "/explore" }, { "event_id": "37820", "occurred_at": "2026-10-31 15:29:45", "event_type": "page_view", "user_id": 2049, "path": "/alerts" }, { "event_id": "499559", "occurred_at": "2026-10-31 15:25:23", "event_type": "page_view", "user_id": 1891, "path": "/billing" }, { "event_id": "443249", "occurred_at": "2026-10-31 15:21:51", "event_type": "page_view", "user_id": 1889, "path": "/dashboard" }, { "event_id": "413561", "occurred_at": "2026-10-31 15:04:21", "event_type": "page_view", "user_id": 1960, "path": "/team" }, { "event_id": "625421", "occurred_at": "2026-10-31 14:55:00", "event_type": "api_call", "user_id": 1905, "path": "/api/v1/users" }, { "event_id": "498146", "occurred_at": "2026-10-31 14:51:04", "event_type": "api_call", "user_id": 2049, "path": "/api/v1/events" }, { "event_id": "141196", "occurred_at": "2026-10-31 14:40:21", "event_type": "api_call", "user_id": 1889, "path": "/api/v1/export" }, { "event_id": "607155", "occurred_at": "2026-10-31 14:39:35", "event_type": "api_call", "user_id": 1963, "path": "/api/v1/users" }, { "event_id": "349393", "occurred_at": "2026-10-31 14:33:08", "event_type": "page_view", "user_id": 1935, "path": "/billing" }, { "event_id": "246017", "occurred_at": "2026-10-31 14:29:26", "event_type": "api_call", "user_id": 1879, "path": "/api/v1/users" }, { "event_id": "185406", "occurred_at": "2026-10-31 14:27:51", "event_type": "api_call", "user_id": 1978, "path": "/api/v1/users" }, { "event_id": "797027", "occurred_at": "2026-10-31 14:17:54", "event_type": "api_call", "user_id": 1909, "path": "/api/v1/query" }, { "event_id": "749079", "occurred_at": "2026-10-31 14:01:44", "event_type": "page_view", "user_id": 1838, "path": "/alerts" }, { "event_id": "472087", "occurred_at": "2026-10-31 13:58:14", "event_type": "api_call", "user_id": 1949, "path": "/api/v1/users" }, { "event_id": "734706", "occurred_at": "2026-10-31 13:56:21", "event_type": "api_call", "user_id": 1905, "path": "/api/v1/events" }, { "event_id": "949525", "occurred_at": "2026-10-31 13:55:40", "event_type": "api_call", "user_id": 1831, "path": "/api/v1/events" }, { "event_id": "575126", "occurred_at": "2026-10-31 13:55:10", "event_type": "export", "user_id": 1903, "path": "/export" }, { "event_id": "388478", "occurred_at": "2026-10-31 13:50:10", "event_type": "login", "user_id": 2042, "path": "/login" }, { "event_id": "433315", "occurred_at": "2026-10-31 13:47:07", "event_type": "api_call", "user_id": 2028, "path": "/api/v1/query" }, { "event_id": "387451", "occurred_at": "2026-10-31 13:47:06", "event_type": "page_view", "user_id": 1903, "path": "/explore" }, { "event_id": "228165", "occurred_at": "2026-10-31 13:23:33", "event_type": "api_call", "user_id": 1828, "path": "/api/v1/users" }, { "event_id": "249186", "occurred_at": "2026-10-31 13:18:10", "event_type": "page_view", "user_id": 1957, "path": "/settings" }, { "event_id": "767466", "occurred_at": "2026-10-31 13:16:42", "event_type": "page_view", "user_id": 2032, "path": "/alerts" }, { "event_id": "81308", "occurred_at": "2026-10-31 13:15:41", "event_type": "page_view", "user_id": 1933, "path": "/settings" }, { "event_id": "116230", "occurred_at": "2026-10-31 13:11:01", "event_type": "page_view", "user_id": 1905, "path": "/billing" }, { "event_id": "284524", "occurred_at": "2026-10-31 13:09:46", "event_type": "page_view", "user_id": 1905, "path": "/alerts" }, { "event_id": "66466", "occurred_at": "2026-10-31 13:05:00", "event_type": "page_view", "user_id": 1970, "path": "/explore" }, { "event_id": "665309", "occurred_at": "2026-10-31 12:58:25", "event_type": "page_view", "user_id": 1848, "path": "/explore" }, { "event_id": "815771", "occurred_at": "2026-10-31 12:55:18", "event_type": "query_run", "user_id": 1831, "path": "/explore" }, { "event_id": "507931", "occurred_at": "2026-10-31 12:53:29", "event_type": "query_run", "user_id": 1933, "path": "/explore" }, { "event_id": "500073", "occurred_at": "2026-10-31 12:52:56", "event_type": "page_view", "user_id": 1842, "path": "/billing" }, { "event_id": "590537", "occurred_at": "2026-10-31 12:42:29", "event_type": "page_view", "user_id": 1841, "path": "/dashboard" }, { "event_id": "602011", "occurred_at": "2026-10-31 12:23:35", "event_type": "error", "user_id": 2033, "path": "/alerts" }, { "event_id": "861576", "occurred_at": "2026-10-31 12:17:18", "event_type": "page_view", "user_id": 2028, "path": "/settings" }, { "event_id": "508131", "occurred_at": "2026-10-31 12:14:23", "event_type": "api_call", "user_id": 2033, "path": "/api/v1/users" } ], "nextCursor": "eyJvY2N1cnJlZEF0IjoiMjAyNi0xMC0zMSAxMjoxNDoyMyIsImV2ZW50SWQiOiI1MDgxMzEifQ" } ``` ## Related - [Multi-tenant embedded analytics on ClickHouse with signed tokens](https://hypequery.com/examples/multi-tenant-embedded-analytics): Embed tenant-scoped ClickHouse analytics in your product: your server mints a short-lived token, hypequery enforces the tenant on every query. - [A FastAPI analytics API on ClickHouse with tenant-scoped datasets](https://hypequery.com/examples/fastapi-analytics-api): Serve governed ClickHouse analytics from FastAPI: datasets defined once in Python, query endpoints mounted beside your routes, tenant fixed by the token. - [All recipes](https://hypequery.com/recipes) - [/clickhouse-query-builder](https://hypequery.com/clickhouse-query-builder) - [/clickhouse-api](https://hypequery.com/clickhouse-api) - [/clickhouse-typescript](https://hypequery.com/clickhouse-typescript)