# Search logs

Read and filter what an app's containers printed, and the log lines of schedules, functions, event rules, load balancers and CDN distributions.

Six kinds of resource ship log lines to Monitoring:

| `resourceType`     | What its lines are                                                          |
| ------------------ | --------------------------------------------------------------------------- |
| `app` (default)    | What the app's containers write to stdout and stderr                        |
| `schedule`         | A schedule's invocations, one stream per invocation                         |
| `function`         | A function's output                                                         |
| `event_rule`       | An event rule's deliveries, one stream per delivery                         |
| `load_balancer`    | An application load balancer's requests                                     |
| `cdn_distribution` | A CDN distribution's requests (stream `access`) and web ACL matches (`waf`) |

An app's stream is `<service>/stdout` or `<service>/stderr`: a single-image app's service is `app`,
a compose app's the compose service name. The agent on the app's machine ships its lines about every
ten seconds. Replace _app-id_ with the resource's id.

## Prerequisites

- Permission for `monitoring:FilterLogEvents` on `rn:ews::<organization id>:monitoring:*`. Without
  it, **Logs** shows an access-denied notice.
- In the console, the **Logs** page lists application load balancers only with
  `lb:DescribeLoadBalancers` and CDN distributions only with `cdn:ListDistributions`.

## Read the lines

**Console**
1. Open **Monitoring**, then **Logs**. An app's own page has the same view of its lines under its
   **Logs** tab, with at most 200 lines.
2. Choose the source: an app, a load balancer or a CDN distribution.
3. (Optional) Type in **Lines containing…** to show only lines that contain the text, in any case.
4. Choose a range: **1 hour**, **6 hours**, **24 hours** or **7 days**.

The newest line is at the bottom, and new lines keep arriving while the view is live; **Pause**
stops that. The view shows at most 1000 lines. Times are UTC.

**CLI**
```sh
dcloud monitoring logs filter --resource-id app-id --contains error --limit 500
```

**API**
```http
GET /monitoring/api/logs?resourceId=app-id&contains=error&limit=500
```

For another kind of resource, add `resourceType`, for example `resourceType=function`.

## Result

The API answers `{ start, end, events }`: up to `limit` lines (200 by default, at most 1000),
newest first, each with its `stream`, `ts` and `message`. `start` defaults to a day before `end`,
and `end` to now. `contains` matches regardless of case, and `%` and `_` in it are taken literally;
`stream` matches one stream exactly. An id your organization does not have answers no lines.

## Limits

- Lines are kept 7 days.
- An app's line longer than 16 KiB is cut and ends in `…[truncated]`.
- When an app prints faster than its machine can ship, the agent keeps at most 5000 lines waiting
  and drops the rest; the app's activity says how many.
- The agent ships what containers print from the moment it starts; earlier output stays on the
  machine.

## Related

- [How Monitoring collects data](how-monitoring-collects-data.md)
- `monitoring.logs.filter` · `dcloud monitoring logs filter`
