From 04d8f5608615eb017ad41a4df0d632fbe0c9258f Mon Sep 17 00:00:00 2001
From: Sam Stoelinga <sammiestoel@gmail.com>
Date: Mon, 28 Oct 2024 22:54:46 -0700
Subject: [PATCH] improve caching docs (#295)

Based on feedback from user trying it out.
---
 docs/how-to/cache-models-with-aws-efs.md      |  4 +++-
 .../how-to/cache-models-with-gcp-filestore.md | 19 +++++++++++++++++++
 2 files changed, 22 insertions(+), 1 deletion(-)

diff --git a/docs/how-to/cache-models-with-aws-efs.md b/docs/how-to/cache-models-with-aws-efs.md
index 27396f5c..582381cd 100644
--- a/docs/how-to/cache-models-with-aws-efs.md
+++ b/docs/how-to/cache-models-with-aws-efs.md
@@ -118,7 +118,7 @@ Make sure to set `file_system_id` match the EFS file system ID created in the fi
 
 ## 3. Configure KubeAI with the EFS cache profile
 
-You can skip this step if you've already installed KubeAI using the [EKS Helm values file: values-eks.yaml](https://github.com/substratusai/kubeai/blob/main/charts/kubeai/values-eks.yaml) file.
+You can skip this step if you've already installed KubeAI using the [EKS Helm values file: values-eks.yaml](https://github.com/substratusai/kubeai/blob/main/charts/kubeai/values-eks.yaml).
 
 Configure KubeAI with the `efs-dynamic` cache profile.
 ```bash
@@ -133,6 +133,8 @@ cacheProfiles:
       persistentVolumeName: "efs-pv"
 EOF
 ```
+You can also manually provision a PersistentVolume using EFS static mode.
+The efs-static profile is used for this purpose.
 
 ## 4. Configure a model to use the EFS cache
 
diff --git a/docs/how-to/cache-models-with-gcp-filestore.md b/docs/how-to/cache-models-with-gcp-filestore.md
index c0d11078..bc5f3c44 100644
--- a/docs/how-to/cache-models-with-gcp-filestore.md
+++ b/docs/how-to/cache-models-with-gcp-filestore.md
@@ -13,6 +13,25 @@ Ensure that the Filestore API is enabled.
 gcloud services enable file.googleapis.com
 ```
 
+## 1. Configure KubeAI with Caching Profile
+
+You can skip this step if you've already installed KubeAI using the [GKE Helm values file: values-gke.yaml](https://github.com/substratusai/kubeai/blob/main/charts/kubeai/values-gke.yaml).
+
+Configure KubeAI with the Filestore cache profiles.
+```bash
+helm upgrade --install kubeai kubeai/kubeai \
+  --reuse-values -f - <<EOF
+cacheProfiles:
+  standard-filestore:
+    sharedFilesystem:
+      storageClassName: "standard-rwx"
+  premium-filestore:
+    sharedFilesystem:
+      storageClassName: "premium-rwx"
+EOF
+```
+
+## 2. Deploy a model that uses the Filestore Caching Profile
 Apply a Model with the cache profile set to `standard-filestore` (defined in the reference [GKE Helm values file](https://github.com/substratusai/kubeai/blob/main/charts/kubeai/values-gke.yaml)).
 
 <details markdown="1">