diff --git a/content/en/containers/kubernetes/csi_driver.md b/content/en/containers/kubernetes/csi_driver.md
index 6985b0f9af7..f15753a2486 100644
--- a/content/en/containers/kubernetes/csi_driver.md
+++ b/content/en/containers/kubernetes/csi_driver.md
@@ -64,7 +64,124 @@ To install and activate the Datadog CSI Driver, set `datadog.csi.enabled` to `tr
{{% tab "Datadog Operator" %}}
-If the Datadog Agent is deployed using the Datadog Operator, you must install the Datadog CSI Driver Helm chart before you activate Datadog CSI in the Datadog Agent.
+
+Operator-managed CSI Driver installation requires Datadog Operator
v1.26.0 or later. If you are using an earlier version, see
Legacy Helm-based installation below.
+
+
+Starting with Datadog Operator v1.26.0, the Operator can install and manage the CSI Driver on your behalf. Set `csi.enabled` to `true` in your `DatadogAgent` resource, and the Operator automatically creates a `DatadogCSIDriver` custom resource to deploy the driver:
+
+```yaml
+apiVersion: datadoghq.com/v2alpha1
+kind: DatadogAgent
+metadata:
+ name: datadog
+spec:
+ global:
+ credentials:
+ apiSecret:
+ secretName: datadog-secret
+ keyName: api-key
+ csi:
+ enabled: true
+```
+
+By default, `csi.autoManage` is `true`, which means the Operator owns the lifecycle of the `DatadogCSIDriver` resource. You can optionally customize scheduling through the `DatadogAgent` spec:
+
+```yaml
+ csi:
+ enabled: true
+ nodeSelector:
+ kubernetes.io/os: linux
+ tolerations:
+ - key: node-role.kubernetes.io/control-plane
+ effect: NoSchedule
+```
+
+{{% collapse-content title="Advanced setup with DatadogCSIDriver CR" level="h4" %}}
+
+If you need configuration options not exposed through the `DatadogAgent` spec (for example, custom images, container resource limits, update strategies, or per-container environment variables), you can manage the `DatadogCSIDriver` custom resource directly:
+
+1. **Disable automatic management** by setting `csi.autoManage` to `false` in your `DatadogAgent` resource:
+
+ ```yaml
+ apiVersion: datadoghq.com/v2alpha1
+ kind: DatadogAgent
+ metadata:
+ name: datadog
+ spec:
+ global:
+ csi:
+ enabled: true
+ autoManage: false
+ ```
+
+2. **Create your own `DatadogCSIDriver` resource** with the desired customizations:
+
+ ```yaml
+ apiVersion: datadoghq.com/v1alpha1
+ kind: DatadogCSIDriver
+ metadata:
+ name: datadog-agent
+ spec:
+ csiDriverImage:
+ tag: "1.1.0"
+ registrarImage:
+ tag: v2.0.0
+ apmSocketPath: /var/run/datadog/apm.socket
+ dsdSocketPath: /var/run/datadog/dsd.socket
+ override:
+ labels:
+ team: platform
+ annotations:
+ custom.example.com/env: production
+ tolerations:
+ - key: node-role.kubernetes.io/control-plane
+ effect: NoSchedule
+ nodeSelector:
+ kubernetes.io/os: linux
+ priorityClassName: system-node-critical
+ updateStrategy:
+ type: RollingUpdate
+ rollingUpdate:
+ maxUnavailable: "5%"
+ containers:
+ csi-node-driver:
+ env:
+ - name: DD_APM_ENABLED
+ value: "false"
+ resources:
+ requests:
+ cpu: 100m
+ memory: 128Mi
+ limits:
+ cpu: 200m
+ memory: 256Mi
+ csi-node-driver-registrar:
+ resources:
+ requests:
+ cpu: 50m
+ memory: 64Mi
+ limits:
+ cpu: 100m
+ memory: 128Mi
+ ```
+
+ See the full [DatadogCSIDriver CRD reference](https://github.com/DataDog/datadog-operator/blob/main/bundle/manifests/datadoghq.com_datadogcsidrivers.yaml) for all available fields.
+
+{{% /collapse-content %}}
+
+
+
Migrating from Helm-based CSI Driver installation
+
If you previously installed the CSI Driver with the standalone Helm chart, Datadog recommends migrating to Operator-managed installation. Choose one of the following approaches:
+
+- Let the Operator manage the CSI Driver: Uninstall the Helm chart (
helm uninstall datadog-csi-driver) and keep the default values for csi.enabled and csi.autoManage. The Operator automatically creates a new DatadogCSIDriver resource and deploys the driver.
+- Keep managing the CSI Driver with Helm: No action is required. The Operator detects the existing
k8s.csi.datadoghq.com CSIDriver and defers to it, regardless of the csi.autoManage value. This backward-compatibility guarantee means your Helm-managed driver continues to run without interference. You can optionally set csi.autoManage: false to make this intent explicit in your configuration.
+
+
+
+{{% collapse-content title="Legacy Helm-based installation (Operator < v1.26.0)" level="h4" id="legacy-helm-based-installation" %}}
+
+If the Datadog Agent is deployed using a Datadog Operator version earlier than v1.26.0, you must install the Datadog CSI Driver Helm chart separately before you activate Datadog CSI in the Datadog Agent.
1. **Add the Datadog CSI Helm repository.**
@@ -72,7 +189,6 @@ If the Datadog Agent is deployed using the Datadog Operator, you must install th
```shell
helm repo add datadog-csi-driver https://helm.datadoghq.com
helm repo update
-
```
2. **Install the Datadog CSI Driver Helm chart.**
@@ -85,7 +201,7 @@ If the Datadog Agent is deployed using the Datadog Operator, you must install th
3. **Activate Datadog CSI in your `DatadogAgent` resource.**
- ```
+ ```yaml
apiVersion: datadoghq.com/v2alpha1
kind: DatadogAgent
metadata:
@@ -100,6 +216,8 @@ If the Datadog Agent is deployed using the Datadog Operator, you must install th
enabled: true
```
+{{% /collapse-content %}}
+
{{% /tab %}}
{{% tab "DaemonSet" %}}