Skip to content
Merged
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
Original file line number Diff line number Diff line change
Expand Up @@ -40,12 +40,12 @@ You can create a pipeline with one of the following methods:
{{< tabs >}}
{{% tab "Logs" %}}

1. Navigate to [Observability Pipelines][7].
1. Select a [template][4] based on your use case.
1. Select and set up your [source][1].
1. Add [processors][2] to transform, redact, and enrich your log data.
1. Navigate to [Observability Pipelines][1].
1. Select a [template][2] based on your use case.
1. Select and set up your [source][3].
1. Add [processors][4] to transform, redact, and enrich your log data.
- If you want to copy a processor, click the copy icon for that processor and then use `command-v` to paste it.
1. Select and set up [destinations][3] for your processed logs.
1. Select and set up [destinations][5] for your processed logs.

### Add or remove components

Expand All @@ -56,7 +56,7 @@ You can create a pipeline with one of the following methods:
If you want to add another group of processors for a destination:
1. Click the plus sign (**+**) at the bottom of the existing processor group.
1. Click the name of the processor group to update it.
1. Optionally, enter a group filter. See [Search Syntax][11] for more information.
1. Optionally, enter a group filter. See [Search Syntax][6] for more information.
1. Click **Add** to add processors to the group.
1. If you want to copy all processors in a group and paste them into the same processor group or a different group:
1. Click the three dots on the processor group.
Expand Down Expand Up @@ -92,17 +92,12 @@ To delete a destination, click on the pencil icon to the top right of the destin
- You can add a total of three destinations for a pipeline.
- A specific destination can only be added once. For example, you cannot add multiple Splunk HEC destinations.

[1]: /observability_pipelines/sources/
[2]: /observability_pipelines/processors/
[3]: /observability_pipelines/destinations/
[4]: /observability_pipelines/configuration/explore_templates/
[5]: /observability_pipelines/configuration/update_existing_pipelines/
[6]: /observability_pipelines/configuration/install_the_worker/
[7]: https://app.datadoghq.com/observability-pipelines
[8]: /monitors/types/metric/
[9]: /observability_pipelines/guide/environment_variables/
[10]: /observability_pipelines/configuration/install_the_worker/advanced_worker_configurations/#bootstrap-options
[11]: /observability_pipelines/search_syntax/logs/
[1]: https://app.datadoghq.com/observability-pipelines
[2]: /observability_pipelines/configuration/explore_templates/
[3]: /observability_pipelines/sources/
[4]: /observability_pipelines/processors/
[5]: /observability_pipelines/destinations/
[6]: /observability_pipelines/search_syntax/logs/

{{% /tab %}}
{{% tab "Metrics" %}}
Expand Down Expand Up @@ -143,23 +138,16 @@ If you want to add another group of processors for a destination:

### Install the Worker and deploy the pipeline

After you have set up your source, processors, and destinations:
After you have set up your source, processors, and destinations, click **Next: Install**. See [Install the Worker][12] for instructions on how to install the Worker for your platform. See [Advanced Worker Configurations][5] for bootstrapping options.

1. Click **Next: Install**.
1. Select the platform on which you want to install the Worker.
1. Enter the [environment variables][9] for your sources and destinations, if applicable.
1. Follow the instructions on installing the Worker for your platform. The command provided in the UI to install the Worker has the relevant environment variables populated.
- See [Install the Worker][6] for more information.
- **Note**: If you are using a proxy, see the `proxy` option in [Bootstrap options][10].
1. Enable out-of-the-box monitors for your pipeline.
1. Navigate to the [Pipelines][7] page and find your pipeline.
1. Click **Enable monitors** in the **Monitors** column for your pipeline.
1. Click **Start** to set up a monitor for one of the suggested use cases.<br>
- The metric monitor is configured based on the selected use case. You can update the configuration to further customize it. See the [Metric monitor documentation][8] for more information.
If you want to make changes to your pipeline after you have deployed it, see [Update Existing Pipelines][11].

After you have set up your pipeline, see [Update Existing Pipelines][11] if you want to make any changes to it.
### Enable out-of-the-box monitors for your pipeline

See [Advanced Worker Configurations][5] for bootstrapping options.
1. Navigate to the [Pipelines][4] page and find your pipeline.
1. Click **Enable monitors** in the **Monitors** column for your pipeline.
1. Click **Start** to set up a monitor for one of the suggested use cases.<br>
- The metric monitor is configured based on the selected use case. You can update the configuration to further customize it. See the [Metric monitor documentation][13] for more information.

## Set up a pipeline with the API

Expand Down Expand Up @@ -228,4 +216,6 @@ To delete a pipeline in the UI:
[8]: /api/latest/observability-pipelines/#update-a-pipeline
[9]: /observability_pipelines/guide/environment_variables/
[10]: https://registry.terraform.io/providers/DataDog/datadog/latest/docs
[11]: /observability_pipelines/configuration/update_existing_pipelines/?
[11]: /observability_pipelines/configuration/update_existing_pipelines/?
[12]: /observability_pipelines/configuration/install_the_worker/
[13]: /monitors/types/metric/
Loading