OpenCost integration

Use the OpenCost Advanced Integration to connect Yarken to an OpenCost API service and retrieve supported OpenCost data through a Yarken-managed integration.

Yarken deploys and maintains the Advanced Integration for your environment. Your organization provides an accessible OpenCost API URL, authorizes the connection, and validates the resulting integration runs and data.

Supported data source

This integration connects to an OpenCost API service. OpenCost provides APIs for reporting Kubernetes resource allocation and infrastructure costs. The exact data sets, attributes, and aggregation rules ingested by Yarken depend on the supported Yarken OpenCost integration configuration.

For background about OpenCost's upstream APIs, see OpenCost API documentation.

What the integration provides

• An Advanced Connection to the OpenCost API service.

• A Yarken-managed integration that retrieves the OpenCost data supported by your deployment.

• An Advanced pipeline for running and monitoring the integration.

• OpenCost data handled through its own Yarken schema rather than the AI Consumption schema.

Where the data appears

After the OpenCost Advanced pipeline loads data, use Analytics > Consumption Data and the Kubernetes Consumption field group to report on Kubernetes cost and resource consumption. This data is separate from AI Usage Data and Analytics > AI Consumption.

To create a Kubernetes Consumption report:

1. Go to Analytics > Consumption Data.

2. Expand Kubernetes Consumption in the Fields panel.

   

3. Select or drag the required fields into the Data Table visualization.

4. Apply a Timeline filter, such as Month, when required.

5. Select Run Report to display the results.

Kubernetes Consumption fields available for reporting

These are the fields visible in the supplied Yarken screenshot. The 34-field OpenCost connector output is not identical to the fields available for reporting. For example, Workload Cost and Idle Cost are reporting measures rather than fields with those names in the connector output.

Roles and responsibilities

Before you begin

Prepare the OpenCost API endpoint

Your Kubernetes or platform administrator must supply one API URL per cluster. OpenCost must be running with Prometheus (or a compatible metrics store) and a configured cloud-billing or custom pricing source. The connection uses the API service, not the dashboard UI.

• Find the API service (for example, kubectl get svc -n opencost). The API normally listens on port 9003; the dashboard commonly uses 9090.

• Expose the API service over HTTPS and preserve its full base path (for example, https://opencost.example.com/api). Both GET /allocation (required) and GET /assets (enrichment) should be reachable.

• Allowlist the integration platform's Workato outbound IP addresses, as advised for your deployment. No API key, token, basic auth, or SSO credential is sent by this connector; do not put an authentication/login proxy in front of the endpoint.

• Ensure the proxy supports large JSON responses and requests lasting at least a minute for busy clusters.

• Confirm metrics retention covers the historical dates you need, then verify the endpoint from outside your network, not only from inside the Kubernetes cluster.

• Send Yarken the URL and associated cluster name, pricing source, Prometheus retention period, and endpoint-test results securely. Cluster identity is not included in the connector output rows.

Connection check — expect HTTP 200 and Content-Type: application/json:

curl -i "https://opencost.example.com/api/allocation?window=1d"
curl -i "https://opencost.example.com/api/assets?window=1d"

The OpenCost UI can return HTTP 200 with HTML. If the response is HTML instead of JSON, use the API service address. A 401/403 or login page indicates unsupported proxy authentication.

Authorize the OpenCost connection

1. Sign in to Yarken as an administrator.

2. Go to Admin > Pipelines > Connections.

3. Select the ADVANCED tab.

4. Open the OpenCost connection provided for your environment.

   

5. Enter the full OpenCost API service address in API URL.

6. Select Connect if the connection has not yet been authorized.

7. Confirm that the connection displays Connected.

OpenCost Advanced Connection: The UI displays a masked API URL and a Connected status once the connection is authorized.

When the connection is already established, Yarken displays Connected and provides a Disconnect action. Do not disconnect an active integration unless you intend to interrupt its access.

For the general authorization procedure, see Authorize Advanced Connections.

How the integration works

1. The OpenCost API service provides source data.

2. The OpenCost Advanced Connection allows Yarken to reach the configured API service.

3. The Yarken-managed Advanced Integration retrieves and processes the supported source data.

4. The Advanced pipeline runs the integration and records execution activity.

5. The data is processed using the OpenCost-specific Yarken schema.

You do not configure the underlying integration logic or define an Advanced Integration recipe yourself.

Source APIs and processing behavior

The connector requires window_start and window_end in RFC3339 UTC (end exclusive). It calls /allocation with the following fixed parameters. Idle allocations are not spread across workloads.

aggregate=namespace,pod,container,controller,controllerKind,node,service
step=1d
resolution=1h
includeIdle=true
idle=true
shareIdle=false
accumulate=false

The connector flattens every allocation entry in every daily step into an output row. It does not aggregate, deduplicate, or filter these rows. /assets is used only to look up Node provider information. The connector tries the root endpoint first, then /model if the root returns 404 or HTML, but does not retry other API failures.

Data available and ingested

The OpenCost connector returns records[] containing 34 flat output fields. The expandable tables below describe the confirmed OpenCost source-to-connector mappings. They do not establish that every connector field is available as an individual field in Analytics. See the Kubernetes Consumption reporting table above for the fields confirmed in the current user interface.

Data interpretation and limitations

• OpenCost cost values depend on configured pricing. Without cloud billing or custom pricing, they may be zero or estimates.

• Idle cost remains a separate __idle__ row; include idle allocation when reconciling total cost.

• Missing or nonnumeric usage and cost values become null, except total_cost, which defaults to 0.0. Empty labels become null; present labels are serialized JSON text.

• The connector output does not contain a cluster field. Identify clusters separately by connection or another implementation-approved method.

• No raw asset costs, asset_type, or asset_key are returned; /assets enriches allocation rows with matching Node identity only.

• Provider and project may be null when /assets fails or an allocation does not match a Node. Idle rows commonly have no node.

• Fixed aggregation does not include label:<key> groupings; labels may therefore be missing.

• History is limited by metrics retention. The connector does not implement paging or automatic retries for timeouts/5xx; use shorter requested windows for large clusters.

Run and monitor the Advanced pipeline

After the OpenCost connection shows Connected:

1. Go to Admin > Pipelines > Pipelines.

2. Select the ADVANCED tab.

3. Locate the OpenCost pipeline enabled for your environment.

4. Select Run when you need to start an immediate execution.

5. Open the pipeline to review its latest status, run history, and any available error details.

For activation and other supported controls, see Manage Advanced pipelines.

Validate the integration

After the first successful run:

• Confirm the OpenCost Advanced Connection shows Connected.

• Confirm the configured API URL points to the OpenCost API service.

• Verify that the corresponding Advanced pipeline completed successfully.

• Review the run history for reported errors or missing source data.

• Go to Analytics > Consumption Data, expand Kubernetes Consumption, choose relevant fields such as Namespace, Pod, Total Cost, Is Idle, and Window Start, then select Run Report. Compare records with OpenCost for the same UTC reporting period.

Troubleshooting

Security and access considerations

• Use an OpenCost API endpoint approved for your organization's integration and network-access policies.

• Do not expose sensitive internal hostnames, network details, or credentials in public screenshots or support messages.

• Coordinate endpoint changes with your OpenCost administrator and update the Advanced Connection when necessary.

• After changing the API URL, validate the connection and run the Advanced pipeline again.

Related content

• Advanced integrations

• Authorize Advanced Connections

• Manage Advanced pipelines

• Troubleshoot Advanced Integrations