# Explore and the query API

> Graph metrics and search logs from the last 7 days in Explore, or with POST /api/v1/query.

**Explore** is the app's home page. Pick **Logs** or **Metrics**, a time range (15 minutes to 7 days, or custom) and filters; Pulse draws the graph or the log table and refreshes every 30 seconds. The query is kept in the page URL, so bookmark or share it.

Data is queryable as soon as it arrives, including rows not yet written to storage, for the **last 7 days** (UTC days, today included). Older days have gone to your bucket; see [Archive](/docs/archive-and-grace/).

## `POST /api/v1/query`

Send an [API key](/docs/api/) with the `read` scope. Times are milliseconds since the epoch.

Logs:

```json
{ "signal": "logs", "from": 1760140800000, "to": 1760144400000,
  "service": "checkout", "minSev": 17, "contains": "timeout", "attrs": { "http.route": "/pay" }, "limit": 200 }
```

```json
{ "signal": "logs", "rows": [ { "time": 1760144391123, "sev": 17, "sevText": "ERROR", "body": "upstream timeout", "service": "checkout", "attrs": {}, "resource": {} } ],
  "histogram": [[1760140800000, 3], [1760141100000, 0]], "scanned": { "files": 2, "bytes": 1830211, "bufferedRows": 412 }, "truncated": false }
```

Metrics:

```json
{ "signal": "metrics", "from": 1760140800000, "to": 1760144400000,
  "name": "http.server.duration", "agg": "p95", "step": 60, "groupBy": ["http.route"] }
```

```json
{ "signal": "metrics", "series": [ { "labels": { "http.route": "/pay" }, "points": [[1760140800000, 182.5]] } ],
  "scanned": { "files": 2, "bytes": 1830211, "bufferedRows": 412 }, "truncated": false }
```

| Field | Logs | Metrics |
| --- | --- | --- |
| `from`, `to` | Required | Required |
| `service`, `minSev`, `contains` | Optional, see [Logs](/docs/logs/) | — |
| `attrs` | Optional exact matches | Optional exact matches |
| `limit` | 1-1,000, default 200, newest first | — |
| `step` | Histogram bucket width in seconds (optional) | Required, at least 10 |
| `name`, `agg`, `groupBy` | — | `name` and `agg` required, see [Metrics](/docs/metrics/) |

## Big queries

A query reads at most 2,000 files or about 1.5 GB of column data, and runs for at most 20 seconds. Past that you get what was read so far with `"truncated": true`: narrow the time range, or filter on `service` or `name`.

Points are `[time, value]` pairs at the start of each step.