> 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/services.md).

# Services

Every service in a workspace on one list, and one page per service with its health, signals, incidents and changes

A service is the thing you usually think in when something breaks: "checkout is slow", "the payments API is throwing errors". Dstl8 gathers everything it knows about each service into one place. The **Services** list shows every service in a workspace with its requests, errors and latency, its log errors, and what is open on it. Each service has its own page with its health over time, its signals on one timeline, its incidents and anomalies, its operations, the services it calls and is called by, its metrics, its logs, and the deploys and rollouts behind it.

Open a workspace and click **Services** in the sidebar, under **Workspace**.

## What counts as a service

A service is a name that your telemetry carries. Dstl8 reads it from each signal and puts the logs, metrics and traces that carry the same name on one row:

* **OpenTelemetry:** the `service.name` resource attribute your SDK or collector sets.
* **Kubernetes logs and metrics:** when a pod's logs or metrics carry no service name, the agent fills one in, in this order: the pod's `resource.opentelemetry.io/service.name` annotation, then its `app.kubernetes.io/instance` label, then its `app.kubernetes.io/name` label, then the name of the workload that runs it (its Deployment, ReplicaSet, StatefulSet, DaemonSet, CronJob or Job).

The **Signals** column shows which of logs, metrics and traces name each service, so you can see at a glance that a service sends traces but no metrics, for example.

{% hint style="warning" %}
**Names must match exactly.** `checkout` and `checkout-svc` are two services in Dstl8, even if they are the same application. If a service shows up twice, or its traces and logs land on different rows, make the names agree at the source: set the same `service.name` in your OpenTelemetry SDK, or add the `resource.opentelemetry.io/service.name` annotation to the pod so its logs use the name its traces use.
{% endhint %}

## The Services list

The list covers the time range in the header (the last hour by default; presets run from 10 minutes to 30 days). It starts with the busiest services first: the most requests, then the most log lines.

| Column          | Shows                                                                                                                   |
| --------------- | ----------------------------------------------------------------------------------------------------------------------- |
| **Service**     | The service name. Click it, or anywhere on the row, to open the service page. Cmd- or Ctrl-click opens it in a new tab. |
| **Signals**     | Icons for logs, metrics and traces, lit when that signal names the service in this window.                              |
| **Activity**    | A sparkline of requests over the window for a traced service, or log lines for one without traces.                      |
| **Req/min**     | Requests per minute, from traces.                                                                                       |
| **Errors**      | The share of requests that failed, in red at 5% or more.                                                                |
| **p95**         | The 95th-percentile request latency.                                                                                    |
| **Log err/min** | Error log lines per minute, in red when there are any.                                                                  |
| **Incidents**   | Open incidents on the service.                                                                                          |
| **Anomalies**   | Open anomalies on the service's metrics and traces.                                                                     |
| **Top anomaly** | The leading open anomaly, in one line.                                                                                  |

A dash means there is nothing to show (no traces for a logs-only service, for example). A question mark means that number could not be read this time; a note under the table says what is missing, and **Refresh** tries again.

Click a column header (Service or any number) to sort by it. On a narrow screen the table drops the less essential columns first (Top anomaly, then Signals, Activity, Log err/min and Anomalies, then p95) so it fits without scrolling sideways. **Service**, **Req/min**, **Errors** and **Incidents** always stay.

### Filters

* **Search services…** narrows the list by name.
* **Has problems** keeps the services that need a look: those with an open incident, an open anomaly, or errors on 5% or more of their requests.
* **Has traces** keeps the services that send traces.

The count beside the filters ("12 of 40 services") tells you how many rows the filters leave. The list shows up to 200 services; services with something open on them are always included, and the count says when some were left out.

## The service page

The page header shows the service name, the time range, and buttons to jump elsewhere:

* **Traces** opens the Traces page filtered to this service (shown when the service sends traces).
* **Logs** opens the Logs page filtered to this service (shown when it sends logs).
* **Open in Canvas** opens a canvas of this service you can rearrange and add to, if your organization has Canvas. See [Canvas](https://docs.controltheory.com/controltheory-documentation/dstl8-docs/canvas.md).

Under the name, the leading open anomaly is shown in one line. Below that, summary tiles give the service's numbers for the window: **Requests / min**, **Errors**, **p95**, **Log errors / min**, **Incidents** and **Anomalies**. A tile turns red when something needs a look.

The sections below appear when the service has the data for them.

### Signals

The service's logs, traces and metrics as lanes on one shared time axis, so you can see whether an error burst in the logs lines up with failing requests or a metric change. See [Signal lanes](https://docs.controltheory.com/controltheory-documentation/dstl8-docs/signals.md#signal-lanes) for how to read them. The metric and trace lanes cover ranges up to a day; on a longer range the card shows the log lane alone and says so.

### Incidents

The open incidents on this service, newest first, in the same rows as the Incidents list. Click one to open it. If the service has more open incidents than are listed, a note says how many more.

### Anomalies

What looks wrong on the service's metrics and traces right now, one row each, labeled as an **Anomaly** (a value off its usual range), a **Forecast** (a value heading for a limit, with the time left), **Silent** (a series that stopped reporting) or **Trend** (a steady drift). Each row opens where to look: **Open in Metrics** for a metric, **Open in Traces** for a trace anomaly. See [Reading metric insights](https://docs.controltheory.com/controltheory-documentation/dstl8-docs/metrics/insights.md) for what each kind means.

### Operations

For a traced service: its operations (the endpoints or span names it serves), busiest first, each with its requests per minute, error share and p95. It shows up to 100 operations, ten at a time with **Load more**. Click a row to open the service's traces.

### Calls in and out

A map of the services that call this one and the services it calls, from its traces, each service with its requests and error share, and each call labeled with its rate, p95 and error share (red when failing). Click another service on the map to open its page. This is often the fastest way to tell whether a problem started here or one hop away.

### Metrics

For a service that sends metrics: the same explorer as the [Metrics page](https://docs.controltheory.com/controltheory-documentation/dstl8-docs/metrics/explore.md), scoped to this service. It opens on the service's busiest metric; pick any other from the list underneath, or group by pod or host to compare replicas. **Open in Metrics** takes the current metric to the full Metrics page, pinned to this service.

### Logs

For a service that sends logs: its log patterns, the most negative first (errors, warnings and poor sentiment alike), and its recent log lines, with a link to see them all on the Logs page.

### Changes

What changed on the service in the window, newest first: deploys, configuration changes, scaling and maintenance recorded on an object of the service's name. Click one to see the details, including the code it changed when the repository is connected to [Code Graph](https://docs.controltheory.com/controltheory-documentation/dstl8-docs/code-graph.md).

Under **From the deployments serving it**, the page also lists the Kubernetes rollouts of the deployments the service runs in, and deploys recorded against those deployments rather than the service, with their versions. A new rollout can take up to five minutes to appear here. To get your CI deploys onto this list, send them as events; see [Sending events](https://docs.controltheory.com/controltheory-documentation/dstl8-docs/events.md#sending-events).

## Walkthrough: a service's error rate went up. What changed?

1. Open **Services** and turn on **Has problems**. Sort by **Errors** if several services are listed.
2. Open the service. The **Errors** tile shows the failing share for the window. If the range is too short to show the start of the problem, widen it in the header.
3. Look at the **Signals** card. The trace lane marks failed requests in red; find where they start, and whether the log lane shows an error burst at the same moment.
4. Check **Anomalies**. A trace anomaly here usually names the failing operation or call; **Open in Traces** takes you to the failing requests.
5. In **Operations**, find the operation with the high error share. If only one operation fails, the problem is in that code path, not the whole service.
6. In **Calls in and out**, look for a red edge. If the calls this service makes to another service are failing, open that service's page and repeat from step 3.
7. Scroll to **Changes**. A deploy or rollout just before the errors started is the first suspect; open it to see the version and, with Code Graph, the code it changed.
8. Ask Möbius. The chat sidebar knows which service and time range you are looking at, so "what changed before the errors started?" works without more context.

To keep the view, click **Open in Canvas** (if your organization has Canvas). The service's canvas includes requests and errors, p95 latency and log errors with deploys drawn as dashed lines, so you can see a deploy and the error rise on one chart.

## Tips

* **Bookmark a service page.** The URL holds the service and the time range, so a link you share opens the same view.
* **Service names elsewhere are links.** A service named on the dashboard's signal headline, the Traces page, an incident's impacted resources, a metric or trace evidence card, or the Sentiment heatmap opens its service page.
* **Logs-only services work too.** A service with only logs still gets a row and a page: log errors, patterns and changes, without the trace sections.
* **"Nothing in this window names" the service?** The service sent nothing in the selected range. Try a longer range, or check that its source is assigned to this workspace (see [Signals and workspaces](https://docs.controltheory.com/controltheory-documentation/dstl8-docs/signals.md)).
* **A service is missing its metrics or traces** but has logs? Either the names differ (see the warning above), or the workspace does not get that signal from the source. See [Signals and workspaces](https://docs.controltheory.com/controltheory-documentation/dstl8-docs/signals.md#choosing-signals-per-workspace).


---

# 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/services.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.
