From 9eba1b5a8ad0292ede3433cb44c9e5f92307991e Mon Sep 17 00:00:00 2001
From: joeyslalom <joeyt@slalom.com>
Date: Tue, 6 Jun 2023 06:47:30 -0700
Subject: [PATCH] Document deploying to GKE (#148)

Describe the permissions and resource I used to leverage workload
identity with a CronJob
---
 docs/deploy-cloud-run.md |   2 +-
 docs/deploy-gke.md       | 104 +++++++++++++++++++++++++++++++++++++++
 2 files changed, 105 insertions(+), 1 deletion(-)
 create mode 100644 docs/deploy-gke.md

diff --git a/docs/deploy-cloud-run.md b/docs/deploy-cloud-run.md
index 98d3be2..c5657b6 100644
--- a/docs/deploy-cloud-run.md
+++ b/docs/deploy-cloud-run.md
@@ -1,6 +1,6 @@
 # Deploy GCR Cleaner to Cloud Run
 
-This document describes how to deploy GCR Cleaner to Cloud Run [Cloud
+This document describes how to deploy GCR Cleaner to [Cloud
 Run][cloud-run] and invoke it via [Cloud Scheduler][cloud-scheduler]. There is
 also a community-supported [Terraform module for gcr-cleaner][tf-module].
 
diff --git a/docs/deploy-gke.md b/docs/deploy-gke.md
new file mode 100644
index 0000000..2ec2471
--- /dev/null
+++ b/docs/deploy-gke.md
@@ -0,0 +1,104 @@
+# Deploy GCR Cleaner to Kubernetes Engine
+
+This document describes how to deploy GCR Cleaner to [Kubernetes Engine][kubernetes-engine] as a [CronJob][cron-job].
+
+
+1. Export your project ID as an environment variable. The rest of this setup
+   assumes this environment variable is set.
+
+   ```sh
+   export PROJECT_ID="my-project"
+   ```
+
+   Note this is your project _ID_, not the project _number_ or _name_.
+
+1. Export the Kubernetes namespace as an environment variable. This is the namespace gcr-cleaner will be deployed.
+
+   ```sh
+   export NAMESPACE="my-namespace"
+   ```
+
+1. Create a service account which will be assigned to the CronJob:
+
+    ```sh
+    gcloud iam service-accounts create "gcr-cleaner" \
+      --project "${PROJECT_ID}" \
+      --display-name "gcr-cleaner"
+    ```
+   
+1. Add the [Artifact Registry Repository Administrator][artifact-repo-admin] role to the service account:
+
+   ```sh
+   gcloud projects add-iam-policy-binding "${PROJECT_ID}" \
+     --member "serviceAccount:gcr-cleaner@${PROJECT_ID}.iam.gserviceaccount.com" \
+     --role "roles/artifactregistry.repoAdmin"
+   ```
+   
+1. (Optional) In order to use the `-recursive`, add [Storage Object Viewer][storage-object-viewer] role to the service account:
+
+   ```sh
+   gcloud projects add-iam-policy-binding "${PROJECT_ID}" \
+     --member "serviceAccount:gcr-cleaner@${PROJECT_ID}.iam.gserviceaccount.com" \
+     --role "roles/storage.objectViewer"
+   ```
+
+1. Configure the service account to use [Workload Identity][workload-identity]:
+
+   ```sh
+   gcloud iam service-accounts add-iam-policy-binding gcr-cleaner@${PROJECT_ID}.iam.gserviceaccount.com \
+     --role roles/iam.workloadIdentityUser \
+     --member "serviceAccount:${PROJECT_ID}.svc.id.goog[${NAMESPACE}/gcr-cleaner]"
+   ```
+
+1. Configure and apply the Kubernetes resources:
+
+```yaml
+apiVersion: v1
+kind: ServiceAccount
+metadata:
+  name: gcr-cleaner
+  namespace: ${NAMESPACE}
+  annotations:
+    iam.gke.io/gcp-service-account: gcr-cleaner@${PROJECT_ID}.iam.gserviceaccount.com
+---
+apiVersion: batch/v1
+kind: CronJob
+metadata:
+  name: gcr-cleaner
+  namespace: ${NAMESPACE}
+spec:
+  concurrencyPolicy: Forbid
+  schedule: "0 0 */1 * *"  # every day
+  jobTemplate:
+    spec:
+      template:
+        metadata:
+          labels:
+            app: gcr-cleaner
+        spec:
+          containers:
+          - name: gcr-cleaner
+            image: "us-docker.pkg.dev/gcr-cleaner/gcr-cleaner/gcr-cleaner-cli"
+            resources:
+              requests:
+                memory: "128Mi"
+                cpu: "250m"
+              limits:
+                memory: "256Mi"
+            args:
+            - -repo=us-central1-docker.pkg.dev/my-project/my-repo
+            - -grace=720h # 30 days
+            - -recursive
+            - -tag-filter-any=.*
+            restartPolicy: Never
+            serviceAccountName: gcr-cleaner
+            nodeSelector:
+              iam.gke.io/gke-metadata-server-enabled: "true"
+```
+
+
+[kubernetes-engine]: https://cloud.google.com/kubernetes-engine
+[cron-job]: https://kubernetes.io/docs/concepts/workloads/controllers/cron-jobs/
+[artifact-repo-admin]: https://cloud.google.com/iam/docs/understanding-roles#artifactregistry.repoAdmin
+[storage-object-viewer]: https://cloud.google.com/iam/docs/understanding-roles#storage.objectViewer
+[workload-identity]: https://cloud.google.com/kubernetes-engine/docs/how-to/workload-identity
\ No newline at end of file