---
isPrivate: true
title: Single Step Instrumentation for Profiling
description: Datadog, the leading service for cloud-scale monitoring.
breadcrumbs: >-
  Docs > Continuous Profiler > Enabling the Profiler > Single Step
  Instrumentation for Profiling
---

> For the complete documentation index, see [llms.txt](https://docs.datadoghq.com/llms.txt).

# Single Step Instrumentation for Profiling

{% alert level="info" %}
Enabling Profiling using Single Step Instrumentation is in Preview.
{% /alert %}

## Overview{% #overview %}

Make sure you are familiar with [Single Step APM Instrumentation](https://docs.datadoghq.com/tracing/trace_collection/automatic_instrumentation/single-step-apm.md) (SSI) before proceeding.

[Continuous Profiler](https://docs.datadoghq.com/profiler.md) can be enabled as part of the SSI setup. This page provides instructions on how to set it up.

## Supported operating systems and environments{% #supported-operating-systems-and-environments %}

SSI is supported only on Linux, for both `x86_64` and `arm64` (ARM v8) architectures. The Preview version of Continuous Profiler with SSI works for host, container, and Kubernetes deployments.

Continuous Profiler with SSI can be enabled for the following languages:

| Language           | SDK version      |
| ------------------ | ---------------- |
| Java               | 1.37.0+          |
| .NET (x86_64 only) | 3.3.1+           |
| Node.js            | 4.39.0+, 5.15.0+ |
| Python             | 3.0.0+           |

Kubernetes deployments require at least Datadog Agent 7.57.0. Host and container deployments can use 7.56.x versions of the Datadog Agent.

## Enable Continuous Profiler with SSI{% #enable-continuous-profiler-with-ssi %}

Continuous Profiler can be enabled as part of the SSI setup by following these steps:

{% tab title="Host and container" %}

1. Go to the [Agent Installation Page](https://app.datadoghq.com/account/settings/agent/latest?platform=overview) and select one of Linux platforms or Docker.
1. Toggle the Enable APM Instrumentation switch. (If there is no switch, the platform is not supported by SSI.) Toggling the switch adds the `DD_APM_INSTRUMENTATION_ENABLED=` environment variable to the installation command, configuring the installed agent to inject the SDK into processes.
1. Copy the installation command into a text editor.
1. Add `DD_PROFILING_ENABLED=auto` as an additional environment variable after `DD_APM_INSTRUMENTATION_ENABLED` in the copied command. This turns on automatic profiler enablement for any new process worth profiling.
1. Proceed with the rest of the installation instructions, using the modified installation command.

{% /tab %}

{% tab title="Kubernetes with Helm Chart" %}

1. Go to the [Agent Installation Page](https://app.datadoghq.com/account/settings/agent/latest?platform=overview) and select Kubernetes, then select Helm Chart.
1. Open the APM dropdown and toggle the Enable APM Instrumentation switch.
1. Add the values below to `datadog-values.yaml` in addition to those indicated by the installation page. The `datadog.profiling.enabled: auto` setting turns on automatic profiler enablement for any new process worth profiling.
1. Proceed with rest of the installation instructions.

```
agents:
  image:
    tag: latest
clusterAgent:
  image:
    tag: latest
datadog:
  profiling:
    enabled: auto
```

If you already have the Datadog Helm chart, ensure it is updated to at least version 3.71.0. Use the "latest" image tag to install a recent Agent version with support for profiling, as the Datadog Helm chart defaults to an older Agent version.
{% /tab %}

{% tab title="Kubernetes with Datadog Operator" %}

1. Go to the [Agent Installation Page](https://app.datadoghq.com/account/settings/agent/latest?platform=overview) and select Kubernetes, then select Operator.
1. Open the APM dropdown and toggle the Enable APM Instrumentation switch.
1. Add the values below to `datadog-values.yaml` in addition to those indicated by the installation page. The `DD_ADMISSION_CONTROLLER_AUTO_INSTRUMENTATION_PROFILING_ENABLED=auto` environment variable on the Cluster Agent turns on automatic profiler enablement for any new process worth profiling.
1. Proceed with rest of the installation instructions.

```
spec:
  override:
    nodeAgent:
      image:
        tag: latest
    clusterAgent:
      image:
        tag: latest
      env:
        - name: DD_ADMISSION_CONTROLLER_AUTO_INSTRUMENTATION_PROFILING_ENABLED
          value: "auto"
```

Use the "latest" image tag to install a recent Agent version with support for profiling, as the Datadog Operator defaults to an older Agent version.
{% /tab %}

## How does profiling work with SSI{% #how-does-profiling-work-with-ssi %}

After the installation, all new processes are executed with the `DD_PROFILING_ENABLED=auto` environment variable. Running processes are not affected. The Datadog library dynamically turns on the profiler for the processes that are good profiling candidates.

The logic for identifying a process as a good candidate varies by language. For Java, all processes are profiled, as Java applications are usually deployed as a single Java process on a host. For Node.js, profiler is only turned on if the application is running for more than 30 seconds and has created at least one tracing span. For Python, when `DD_PROFILING_ENABLED=auto` (the SSI default), profiling is enabled for SSI-instrumented processes.

SSI can also be configured to inject profiling on each and every process by using the value `true` instead of `auto`.

**Note**: Datadog recommends using the `auto` setting to avoid profiling low value processes.

# Reverting{% #reverting %}

The [Single Step APM Instrumentation](https://docs.datadoghq.com/tracing/trace_collection/automatic_instrumentation/single-step-apm.md) page contains instructions for removing all instrumentation. Removing instrumentation also removes profiling.

Additionally, you can disable profiling by taking one of the following steps:

- Repeat the installation instructions using a `false` value in place of `auto`.
- Remove the `DD_PROFILING_ENABLED` setting from the `/etc/environment` file on the host, for host and container deployments.

Finally, you can disable profiling on a per-process basis by explicitly setting `DD_PROFILING_ENABLED=false` on its command line.

## Special considerations for Systemd services{% #special-considerations-for-systemd-services %}

For host installs, the `DD_PROFILING_ENABLED` environment variable is stored in the `/etc/environment` file, where most Linux systems automatically pick it up for all processes. Exceptions to this are systemd services which ignore the file. For applications deployed as systemd services, you need the following line to the application `.service` files:

```
EnvironmentFile=/etc/environment
```

Applications deployed on a host as something other than a systemd service, or container-deployed applications do not need this extra step.