# Delete logs

**DELETE /api/logs**

Delete log rows in bounded batches.

`start_time` / `end_time` carry the same inclusive semantics as `GET /api/logs`: a row is deleted when its `createdAt` falls in `[start_time, end_time]`, at millisecond granularity. Note that this operation removes **every** log row in that window, while `GET /api/logs` returns audit logs only.

Webhook delivery logs (`TriggerHook.*`) share the `logs` table but are read through `GET /api/hooks/{id}/recent-logs` rather than by `GET /api/logs`, so `deletedCount` can exceed the number of rows an export returned. A window covering the last 24 hours also removes the deliveries counted by the execution stats on `GET /api/hooks`.

`end_time` is required, so a bare `DELETE /api/logs` never means "delete everything". `end_time` must not be in the future and is rejected with `400` otherwise; a common cause is passing microseconds or nanoseconds instead of milliseconds, or a miscomputed timezone offset. Passing seconds instead yields a boundary in the past, which deletes nothing rather than deleting too much.

This operation is irreversible and reachable by any Management API token. Each call deletes at most `limit` rows in a single statement and reports `hasMore`. Keep both bounds fixed across calls and repeat while `hasMore` is `true`; recomputing a boundary between calls would skip rows. Run one cleanup at a time: a concurrent purge can make `hasMore` under-report.

### Performance

`DELETE` marks rows dead and leaves index cleanup to autovacuum, so writes are dominated by heap tuples and WAL. The statement is matched on `ctid`, so the outer side is a Tid Scan over exactly `limit` rows and the inner select reads those row ids from `logs__created_at_id`; expect the cost of a call to track `limit` rather than the size of the table. Measure with `EXPLAIN (ANALYZE, BUFFERS)` on your own data before running a large purge.

Freed space is reused by later inserts but the table file does not shrink, so disk usage will not drop immediately after a purge.


## Servers
- Logto endpoint address.: https://[tenant_id].logto.app (Logto endpoint address.)


## Authentication methods
- O auth2


## Parameters


### Query parameters

- **end_time** (string)
  Inclusive upper bound on `createdAt`, in unix milliseconds, with the same semantics as `GET /api/logs`. Only log rows with `createdAt <= end_time` are deleted. Returns `400` if the value is in the future.

- **start_time** (string)
  Inclusive lower bound on `createdAt`, in unix milliseconds, with the same semantics as `GET /api/logs`. Only log rows with `createdAt >= start_time` are deleted. Omit to purge everything up to `end_time`. Returns `400` if it is after `end_time`.

- **limit** (string)
  Maximum rows deleted by this call. Default `1000`, maximum `10000`. Returns `400` for values outside that range.



## Responses
### 200
Summary of the rows deleted by this call. `deletedCount` is the number of log rows deleted, and `deletedRange` is an optional `{ oldest, newest }` object holding the unix-millisecond timestamps of the deleted rows; it is omitted when nothing was deleted. `hasMore` is `true` when `deletedCount` equals `limit`, meaning this batch was full rather than that more rows definitely exist, so the caller should repeat the request with the same bounds.

#### Body: application/json (object)
- **deletedCount** (number)

- **hasMore** (boolean)

- **deletedRange** (object)


### 400
Returned when `end_time` is missing, empty, negative, outside the safe integer range, in the future, or not a finite number; when `start_time` is malformed or after `end_time`; or when `limit` is outside `1..10000`.

### 401
Unauthorized

### 403
Forbidden


[Powered by Bump.sh](https://bump.sh)