> For the complete documentation index, see [llms.txt](https://docs.controltheory.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.controltheory.com/controltheory-documentation/dstl8-docs/metrics.md).

# Metrics

Metrics distilled into statements — what needs a look, what is heading for a limit — instead of raw points

Dstl8 reads your metrics the way it reads your logs: it learns what each metric usually does and tells you when that changes. Instead of a wall of charts, you get statements like:

> api-server · container.memory.working\_set: rising 4.2%/h for 5h, 3.4σ over baseline, \~2.9h to pod limit

(σ is how far a value is from its usual, measured in the metric's normal spread; see [Reading a headline](https://docs.controltheory.com/controltheory-documentation/dstl8-docs/metrics/insights.md#reading-a-headline).)

Every metric group (one metric for one service, workload, pod or host) gets a **usual range** (which allows for time of day once it has a couple of days of history), a **trend**, and, when something is off, one of four kinds of issue: an **Anomaly**, a **Forecast** of a limit being reached, a metric that went **Silent**, or a steady **Trend**. Möbius uses the same reads when it investigates an incident, and cites them as evidence beside log lines.

{% hint style="info" %}
Metrics is in beta, and your plan must include it (**Organization Settings** → **General** lists it under **Features**). There are no alert rules to write and nothing to tune per metric: the same thresholds apply to every organization.
{% endhint %}

## How metrics reach Dstl8

| Path                               | What it collects                                                                               | Set up on                                                                                                                                        |
| ---------------------------------- | ---------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------ |
| **OpenTelemetry source**           | Metrics your apps or collector send over OTLP/HTTP, on the same endpoint as logs and traces    | [OpenTelemetry (OTel)](https://docs.controltheory.com/controltheory-documentation/dstl8-docs/sources/opentelemetry-otel.md)                      |
| **Docker**                         | Metrics your apps send over OTLP to the agent on the host (ports 4317 and 4318)                | [Docker](https://docs.controltheory.com/controltheory-documentation/dstl8-docs/sources/docker.md)                                                |
| **Kubernetes: Prometheus metrics** | kubelet, cAdvisor, node-exporter and kube-state-metrics, scraped by the agent every 60 seconds | [Kubernetes](https://docs.controltheory.com/controltheory-documentation/dstl8-docs/sources/kubernetes.md), switch **Collect Prometheus metrics** |
| **Kubernetes: Remote Write**       | Series your own Prometheus sends with Remote Write 2.0                                         | the same switch                                                                                                                                  |
| **Kubernetes: OTLP from pods**     | Metrics your pods already push over OTLP to the agent on their node                            | `dstl8 sources metrics <source> --prom-metrics --otlp-metrics` (or `--otlp-metrics` alone without Prometheus metrics)                            |

A Kubernetes source collects **no metrics** until you turn one of these on. See [Collecting metrics](https://docs.controltheory.com/controltheory-documentation/dstl8-docs/metrics/collecting.md) for each path, what it scrapes and the [agent version](https://docs.controltheory.com/controltheory-documentation/dstl8-docs/metrics/collecting.md#agent-versions) it needs, and [Series limits](https://docs.controltheory.com/controltheory-documentation/dstl8-docs/metrics/series-limits.md) for how many series each agent keeps.

## Where you see them

* **Explore → Metrics:** each workspace's Metrics page: summary tiles, a **Needs attention** section (active metric issues and the capacity horizon), an explorer for any metric over time against its usual range, and a list of every metric. See [Exploring Metrics](https://docs.controltheory.com/controltheory-documentation/dstl8-docs/metrics/explore.md). A service's page also has a **Metrics** card for that service.
* **Incidents:** an incident's **Signals** section shows the metrics behind it: now, usually, and the peak, with the limit and any deploy marked.
* **Sources:** a source's page has a **Metric series** card showing how many series it sends against its budget. See [Series limits](https://docs.controltheory.com/controltheory-documentation/dstl8-docs/metrics/series-limits.md).
* **CLI:** `dstl8 metrics top` ranks metric groups by what needs attention; `dstl8 metrics watch <metric>` follows one metric as sparklines. See [Metrics CLI](https://docs.controltheory.com/controltheory-documentation/dstl8-docs/metrics/cli.md).
* **MCP:** your own assistant gets the same metric tools Möbius uses. See [Metrics MCP Tools](https://docs.controltheory.com/controltheory-documentation/dstl8-docs/metrics/mcp-tools.md).

## Which workspace sees which metrics

A workspace sees the metrics of the sources and streams assigned to it, like logs, and only when **Metrics** is among its signals. On Kubernetes, a scraped series is filed under its pod's workload, as that pod's log lines are (node-level series under the node). From an OpenTelemetry source or Docker, metrics are scoped by service (`service.name`). See [Signals and Workspaces](https://docs.controltheory.com/controltheory-documentation/dstl8-docs/signals.md) for how to assign a service and choose signals.

## Getting started

1. Send metrics, or turn a path on: see [Collecting metrics](https://docs.controltheory.com/controltheory-documentation/dstl8-docs/metrics/collecting.md).
2. Make sure the workspace gets them: the source (or its service) is assigned, with **Metrics** among the workspace's signals.
3. Open **Explore → Metrics** a few minutes later. Values and the **All metrics** list fill in first; each metric gets its usual range over its first couple of hours, and the **Needs attention** lists start filling after that.

## Next

* [Collecting metrics](https://docs.controltheory.com/controltheory-documentation/dstl8-docs/metrics/collecting.md): turn on each path
* [Exploring Metrics](https://docs.controltheory.com/controltheory-documentation/dstl8-docs/metrics/explore.md): the Metrics page, step by step
* [Reading metric insights](https://docs.controltheory.com/controltheory-documentation/dstl8-docs/metrics/insights.md): what Anomaly, Forecast, Silent and Trend mean, and how to read a headline
* [Series limits](https://docs.controltheory.com/controltheory-documentation/dstl8-docs/metrics/series-limits.md): the series budget and how to stay under it
* [Metrics CLI](https://docs.controltheory.com/controltheory-documentation/dstl8-docs/metrics/cli.md): `metrics top`, `metrics watch`, `sources metrics`
* [Metrics MCP Tools](https://docs.controltheory.com/controltheory-documentation/dstl8-docs/metrics/mcp-tools.md): the metric tools and the evidence they print


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation by asking a question.

Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters:

```
GET https://docs.controltheory.com/controltheory-documentation/dstl8-docs/metrics.md?ask=<question>&goal=<user_goal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case.

The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.
