Release v0.15.0 (#281)

Dockerfile

@@ -31,7 +31,7 @@ ENTRYPOINT ["./test/continuous/run.sh"]
 
 
 FROM base as deploy
-RUN for cluster in gke mig mig-with-container slurm; do \
+RUN for cluster in gke gke-beta mig mig-with-container slurm; do \
     terraform -chdir="./terraform/modules/cluster/${cluster}" init; done
 COPY scripts ./scripts
 ENTRYPOINT ["./scripts/entrypoint.sh"]

Quickstart.md

@@ -0,0 +1,139 @@
+# Quick start Using Docker container
+
+1. **User authentication**
+
+    For using the Cluster Provisioning Tool(CPT) via docker container to create GPU
+    clusters users need to authenticate with GCP. Authentication can happen in 2 ways
+
+   - User authentication via gcloud and passing the authentication directory to in docker
+   run command.
+
+        ```gcloud
+        gcloud auth login --update-adc
+        ```
+
+      Users can pass the authentication directory to docker run command via `-v
+    ~/.config/gcloud:/root/.config/gcloud` in Linux OS and `-v
+    C:\Users%username%\AppData\Roaming\gcloud:/root/.config/gcloud` in Windows OS.
+
+   - Users can choose to prompt the CPT for authentication. They can simply skip
+   providing the authentication directory in the `docker run` command and CPT will
+   prompt for Authentication.
+
+2. **Project level Access**
+
+    Project Owner or Project Editor access is advised.
+3. **Using docker image**
+
+    - Create `terraform.tfvars` file. The sample `terraform.tfvars` for different
+      cluster types can be found [here](./samples/other/).
+    - Command to create/destroy Clusters. More details on different cluster types can
+      be found [here](./README.md).
+
+      > Linux
+
+      ```Linux
+      docker run -it -v ${PWD}:/root/aiinfra/input \
+      -v ~/.config/gcloud:/root/.config/gcloud \
+      us-docker.pkg.dev/gce-ai-infra/cluster-provision-dev/cluster-provision-image:latest \
+      (create|destroy) (mig|gke|gke-beta|mig-with-container)
+      ```
+
+      > Windows
+
+      ```windows
+      docker run -it -v ${PWD}:/root/aiinfra/input \
+      -v C:\Users%username%\AppData\Roaming\gcloud:/root/.config/gcloud \
+      us-docker.pkg.dev/gce-ai-infra/cluster-provision-dev/cluster-provision-image:latest \
+      (create|destroy) (mig|gke|gke-beta|mig-with-container)
+      ```
+
+# Quick start using Terraform module
+
+1. **User authentication**
+
+    For using the Cluster Provisioning Tool(CPT) via docker container to create GPU
+    clusters users need to authenticate with GCP. Authentication can happen in 2 ways
+
+    ```gcloud
+    gcloud auth login --update-adc
+    ```
+
+1. **Project level Access**
+
+    Project Owner or Project Editor access is advised.
+1. **Using docker image**
+
+    - Create `terraform.tfvars` file. The sample `terraform.tfvars` for different
+      cluster types can be found [here](./samples/other/).
+    - Create `backend.tf` file using the content below. Please make sure the
+      `bucketName/foldername` GCP storage path exists before creating resources. User needs to
+      have `storage object owner access` to create the GPC storage path if it does
+      not exist.
+      > **NOTE** The `foldername` needs to be unique for each GPU cluster. Otherwise
+      > it may corrupt previously existing clusters.
+
+      ```terraform
+      terraform {
+        backend "gcs" {
+          // Please make sure this storage path exists
+          bucket = "bucketName"
+          prefix = "foldername"
+        }
+      }
+      ```
+
+    - Create `main.tf` file. Add a reference to the provisioning tool Terraform
+      module. The reference can be added like below
+      > MIG
+
+      ```terraform
+      module "aiinfra-mig" {
+        source             = "github.com/GoogleCloudPlatform/ai-infra-cluster-provisioning//terraform/modules/cluster/mig"
+      }
+      ```
+
+      > GKE
+
+      ```terraform
+      module "aiinfra-gke" {
+        source             = "github.com/GoogleCloudPlatform/ai-infra-cluster-provisioning//terraform/modules/cluster/gke"
+      }
+      ```
+
+      > SLURM
+
+      ```terraform
+      module "aiinfra-slurm" {
+        source             = "github.com/GoogleCloudPlatform/ai-infra-cluster-provisioning//terraform/modules/cluster/slurm"
+      }
+      ```
+
+    - In the `main.tf` file add a variable block for each corresponding variables in
+      `terraform.tfvars`. Below is an example of `terraform .tfvars` file and the
+      corresponding variable block in the `main.tf` that needs to be added.
+
+      > `main.tf`
+
+      ```terraform
+      variable "project_id" {}
+      variable "resource_prefix" {}
+      variable "target_size" {}
+      variable "zone" {}
+      variable "machine_type" {}
+      ```
+
+      > `terraform.tfvars`
+
+      ```terraform
+      project_id      = "project-id"
+      resource_prefix = "simple-mig"
+      target_size     = 1
+      zone            = "us-central1-a"
+      machine_type    = "n1-standard-1"
+      ```
+
+    - Then run the below terraform commands
+      - `terraform init` to initialize terraform
+      - `terraform apply` to create the GPU cluster
+      - `terraform destroy` to cleanup the GPU cluster