> For the complete documentation index, see [llms.txt](https://developer.harness.io/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://developer.harness.io/resilience-testing/3.0/chaos-testing/probes/k8s-probe.md).

# Kubernetes Probe

K8s Probe verifies the desired state of Kubernetes resources by allowing you to define the resource GVR (group-version-resource) with appropriate filters (field selectors or label selectors). The probe supports three operations:

* **present**: Checks that matching Kubernetes resources exist
* **absent**: Checks that matching Kubernetes resources do not exist
* **delete**: Deletes matching Kubernetes resources

### When to use <a href="#when-to-use" id="when-to-use"></a>

* Verify that pods remain in `Running` state during a fault (use `present` with field selector `status.phase=Running`)
* Confirm that a Kubernetes resource (e.g., PVC, ConfigMap) is cleaned up after a fault (use `absent`)
* Delete specific resources as part of experiment teardown (use `delete`)
* Validate that custom resources or operator-managed resources maintain their expected state under failure conditions

### Steps to configure <a href="#steps-to-configure" id="steps-to-configure"></a>

1. Navigate to **Project Settings** > **Chaos Probes** and click **+ New Probe**
2. Select **Kubernetes Probe** and provide a name, and optionally a description and tags
3. Under **Variables**, define any reusable values you want to reference in probe properties or run properties. For each variable, specify the type (`String` or `Number`), name, value (fixed or runtime input), and whether it's required at runtime.
4. Under **Probe Properties**, configure:

   | Field                         | Description                                                                        |
   | ----------------------------- | ---------------------------------------------------------------------------------- |
   | **Kubernetes Resource Group** | The API group of the resource (leave empty for core resources like pods, services) |
   | **Version**                   | The API version (e.g., `v1`, `v1alpha1`)                                           |
   | **Resource**                  | Kind of resource (e.g., `pods`, `deployments`, `persistentvolumeclaims`)           |
   | **Resource Names** (optional) | Comma-separated resource names to target                                           |
   | **Namespace**                 | The namespace to target (e.g., `default`)                                          |
   | **Field Selector** (optional) | Filter by field values (e.g., `status.phase=Running`)                              |
   | **Label Selector** (optional) | Filter by labels (e.g., `app=nginx`)                                               |
   | **Operation**                 | The operation type: `present`, `absent`, or `delete`                               |
5. Provide the **Run Properties**:

   | Field                          | Description                                     |
   | ------------------------------ | ----------------------------------------------- |
   | **Timeout**                    | Maximum time for probe execution (e.g., `10s`)  |
   | **Interval**                   | Time between successive executions (e.g., `2s`) |
   | **Attempt**                    | Number of retry attempts (e.g., `1`)            |
   | **Polling Interval**           | Time between retries (e.g., `30s`)              |
   | **Initial Delay**              | Delay before first execution (e.g., `5s`)       |
   | **Verbosity**                  | Log detail level                                |
   | **Stop On Failure** (optional) | Stop the experiment if the probe fails          |
6. Click **Create Probe**