From ce2935952f19d4cfbe9077fe091503c605f12621 Mon Sep 17 00:00:00 2001
From: Nate Andrews <nandrews@ibm.com>
Date: Mon, 20 Oct 2025 10:44:34 -0400
Subject: [PATCH] added v3 GKE Autopilot install guide

---
 .../gke-autopilot-install.md                  | 99 ++-----------------
 1 file changed, 7 insertions(+), 92 deletions(-)

diff --git a/install-and-configure/install/provider-installations/gke-autopilot-install.md b/install-and-configure/install/provider-installations/gke-autopilot-install.md
index fa5445dfe..252b056d0 100644
--- a/install-and-configure/install/provider-installations/gke-autopilot-install.md
+++ b/install-and-configure/install/provider-installations/gke-autopilot-install.md
@@ -2,104 +2,19 @@
 
 ## Installing Kubecost for GKE Autopilot via Helm
 
-Installing Kubecost on a GKE Autopilot cluster is similar to other cloud providers with Helm v3.1+, with a few changes. Autopilot requires the use of [Google Managed Prometheus](/install-and-configure/advanced-configuration/custom-prom/gcp-gmp-integration.md) service, which generates additional costs within your Google Cloud account.
+Installing Kubecost on a GKE Autopilot cluster is similar to other cloud providers with Helm v3.1+, with a few changes. Autopilot restricts privileged containers by default so the kubecost-network-costs container needs to be disabled.
 
-`helm install kubecost/cost-analyzer -n kubecost -f values.yaml`
-
-Your _values.yaml_ files must contain the below parameters. Resources are specified for each section of the Kubecost deployment, and Pod Security Policies are disabled.
-
-```yaml
-kubecostProductConfigs:
-  clusterName: "<clusterName>" # used for display in Kubecost UI
-
-kubecostModel:
-  promClusterIDLabel: cluster # warning: usage and efficiency will show as zero without this setting enabled
-  resources:
-    requests:
-      cpu: 500m
-      memory: 512Mi
-
-kubecostFrontend:
-  resources:
-    requests:
-      cpu: 250m
-      memory: 55Mi
-
-global:
-  gmp:
-    enabled: true # If true, kubecost will be configured to use GMP Prometheus image and query from Google Cloud Managed Service for Prometheus.
-    gmpProxy:
-      projectId: <GCP Project ID>
-
-serviceAccount:
-  annotations:
-    iam.gke.io/gcp-service-account: <GCP Service Account with Metrics Writer/Reader Permissions> 
-
-prometheus:
-  server:
-    image:
-      repository: gke.gcr.io/prometheus-engine/prometheus
-      tag: v2.41.0-gmp.4-gke.1
-    global:
-      external_labels:
-        cluster_id: <GKE Autopilot Cluster Name>  # cluster_id should be unique for all clusters and the same value as .kubecostProductConfigs.clusterName
-  podSecurityPolicy:
-    enabled: false
-  configmapReload:
-    prometheus:
-      resources:
-        requests:
-          cpu: 250m
-          memory: 256Mi
-  nodeExporter:
-    enabled: false
-  serviceAccounts:
-    nodeExporter:
-      create: false
-    server:
-      annotations: 
-        iam.gke.io/gcp-service-account: <GCP Service Account with Metrics Writer/Reader Permissions>
-
-networkCosts:
-  enabled: false
-
-podSecurityPolicy:
-  enabled: false​
-
-grafana:
-  rbac:
-    pspEnabled: false
-  resources:
-    requests:
-      cpu: 250m
-      memory: 128Mi
-  sidecar:
-    resources:
-      requests:
-        cpu: 250m
-        memory: 384Mi
-
-```
-
-## Turning on Kubelet/cAdvisor scraping via Google Managed Collector
-
-Open the OperatorConfig on your Autopilot Cluster resource for editing:
 
 ```bash
-kubectl -n gmp-public edit operatorconfig config
+helm upgrade --install kubecost --repo https://kubecost.github.io/kubecost/ kubecost --namespace kubecost --create-namespace --values values.yaml
 ```
 
-Add the following collection section to the resource:
+Your _values.yaml_ files must contain the below parameters.
 
 ```yaml
-apiVersion: monitoring.googleapis.com/v1
-kind: OperatorConfig
-metadata:
-  namespace: gmp-public
-  name: config
-collection:
-  kubeletScraping:
-    interval: 30s
+global:
+  clusterId: "clusterName" # used for display in Kubecost UI
+networkCosts:
+  enabled: false
 ```
 
-Save the file and close the editor. After a short time, the Kubelet metric endpoints will be scraped and the metrics become available for querying in Managed Service for Prometheus.