Skip to content

Commit

Permalink
replace suffix with build_id
Browse files Browse the repository at this point in the history
  • Loading branch information
@hsachdevah
hsachdevah committed Mar 18, 2024
1 parent 05cc3b2 commit 4a6fe4b
Show file tree
Hide file tree
Showing 2 changed files with 45 additions and 47 deletions.
91 changes: 45 additions & 46 deletions cloudbuild.yaml
View file
Original file line number Diff line number Diff line change
Expand Up @@ -59,7 +59,7 @@ steps:
terraform apply \
-var-file=tfvars_tests/standard-gke-public.platform.tfvars \
-var=project_id=$PROJECT_ID \
-var=cluster_name=ml-$SHORT_SHA-$_PR_NUMBER-$_SUFFIX-cluster \
-var=cluster_name=ml-$SHORT_SHA-$_PR_NUMBER-$BUILD_ID-cluster \
-var=autopilot_cluster=$_AUTOPILOT_CLUSTER \
-var=cluster_location=$_REGION \
-auto-approve -no-color -lock=false
Expand All @@ -78,25 +78,25 @@ steps:
# Get kube config
gcloud container clusters get-credentials \
ml-$SHORT_SHA-$_PR_NUMBER-$_SUFFIX-cluster \
ml-$SHORT_SHA-$_PR_NUMBER-$BUILD_ID-cluster \
--location $_REGION \
--project $PROJECT_ID
cd /workspace/applications/ray/
terraform apply \
-var-file=workloads.tfvars \
-var=project_id=$PROJECT_ID \
-var=cluster_name=ml-$SHORT_SHA-$_PR_NUMBER-$_SUFFIX-cluster \
-var=cluster_name=ml-$SHORT_SHA-$_PR_NUMBER-$BUILD_ID-cluster \
-var=cluster_location=$_REGION \
-var=kubernetes_namespace=ml-$SHORT_SHA-$_SUFFIX \
-var=workload_identity_service_account=ray-sa-$SHORT_SHA-$_SUFFIX \
-var=gcs_bucket=gke-aieco-ray-$SHORT_SHA-$_SUFFIX \
-var=kubernetes_namespace=ml-$SHORT_SHA-$BUILD_ID \
-var=workload_identity_service_account=ray-sa-$SHORT_SHA-$BUILD_ID \
-var=gcs_bucket=gke-aieco-ray-$SHORT_SHA-$BUILD_ID \
-auto-approve -no-color -lock=false
echo "pass" > /workspace/user_result.txt
# Make sure pods are running
kubectl wait --all pods -n ml-$SHORT_SHA-$_SUFFIX --for=condition=Ready --timeout=300s
kubectl port-forward -n ml-$SHORT_SHA-$_SUFFIX service/example-cluster-kuberay-head-svc 8265:8265 &
kubectl wait --all pods -n ml-$SHORT_SHA-$BUILD_ID --for=condition=Ready --timeout=300s
kubectl port-forward -n ml-$SHORT_SHA-$BUILD_ID service/example-cluster-kuberay-head-svc 8265:8265 &
# Wait port-forwarding to take its place
sleep 5s
Expand All @@ -118,11 +118,11 @@ steps:
terraform destroy \
-var-file=workloads.tfvars \
-var=project_id=$PROJECT_ID \
-var=cluster_name=ml-$SHORT_SHA-$_PR_NUMBER-$_SUFFIX-cluster \
-var=cluster_name=ml-$SHORT_SHA-$_PR_NUMBER-$BUILD_ID-cluster \
-var=cluster_location=$_REGION \
-var=kubernetes_namespace=ml-$SHORT_SHA-$_SUFFIX \
-var=workload_identity_service_account=ray-sa-$SHORT_SHA-$_SUFFIX \
-var=gcs_bucket=gke-aieco-ray-$SHORT_SHA-$_SUFFIX \
-var=kubernetes_namespace=ml-$SHORT_SHA-$BUILD_ID \
-var=workload_identity_service_account=ray-sa-$SHORT_SHA-$BUILD_ID \
-var=gcs_bucket=gke-aieco-ray-$SHORT_SHA-$BUILD_ID \
-auto-approve -no-color -lock=false
allowFailure: true
Expand All @@ -143,16 +143,16 @@ steps:
terraform apply \
-var-file=workloads-without-iap.example.tfvars \
-var=project_id=$PROJECT_ID \
-var=cluster_name=ml-$SHORT_SHA-$_PR_NUMBER-$_SUFFIX-cluster \
-var=kubernetes_namespace=ml-$SHORT_SHA-$_SUFFIX \
-var=workload_identity_service_account=jupyter-sa-$SHORT_SHA-$_SUFFIX \
-var=gcs_bucket=gke-aieco-jupyter-$SHORT_SHA-$_SUFFIX \
-var=cluster_name=ml-$SHORT_SHA-$_PR_NUMBER-$BUILD_ID-cluster \
-var=kubernetes_namespace=ml-$SHORT_SHA-$BUILD_ID \
-var=workload_identity_service_account=jupyter-sa-$SHORT_SHA-$BUILD_ID \
-var=gcs_bucket=gke-aieco-jupyter-$SHORT_SHA-$BUILD_ID \
-auto-approve -no-color -lock=false
echo "pass" > /workspace/jupyterhub_tf_result.txt
kubectl wait --all pods -n ml-$SHORT_SHA-$_SUFFIX --for=condition=Ready --timeout=300s
kubectl get services -n ml-$SHORT_SHA-$_SUFFIX
kubectl get service proxy-public -n ml-$SHORT_SHA-$_SUFFIX --output jsonpath='{.status.loadBalancer.ingress[0].ip}' > /workspace/jupyterhub_host_url.txt
kubectl wait --all pods -n ml-$SHORT_SHA-$BUILD_ID --for=condition=Ready --timeout=300s
kubectl get services -n ml-$SHORT_SHA-$BUILD_ID
kubectl get service proxy-public -n ml-$SHORT_SHA-$BUILD_ID --output jsonpath='{.status.loadBalancer.ingress[0].ip}' > /workspace/jupyterhub_host_url.txt
echo "HOST URL is " $(cat /workspace/jupyterhub_host_url.txt)
cd /workspace/modules/jupyter/tests
Expand All @@ -172,10 +172,10 @@ steps:
terraform destroy \
-var-file=workloads-without-iap.example.tfvars \
-var=project_id=$PROJECT_ID \
-var=cluster_name=ml-$SHORT_SHA-$_PR_NUMBER-$_SUFFIX-cluster \
-var=kubernetes_namespace=ml-$SHORT_SHA-$_SUFFIX \
-var=workload_identity_service_account=jupyter-sa-$SHORT_SHA-$_SUFFIX \
-var=gcs_bucket=gke-aieco-jupyter-$SHORT_SHA-$_SUFFIX \
-var=cluster_name=ml-$SHORT_SHA-$_PR_NUMBER-$BUILD_ID-cluster \
-var=kubernetes_namespace=ml-$SHORT_SHA-$BUILD_ID \
-var=workload_identity_service_account=jupyter-sa-$SHORT_SHA-$BUILD_ID \
-var=gcs_bucket=gke-aieco-jupyter-$SHORT_SHA-$BUILD_ID \
-auto-approve -no-color -lock=false
allowFailure: true
Expand All @@ -191,7 +191,7 @@ steps:
# Get kube config
gcloud container clusters get-credentials \
ml-$SHORT_SHA-$_PR_NUMBER-$_SUFFIX-cluster \
ml-$SHORT_SHA-$_PR_NUMBER-$BUILD_ID-cluster \
--location $_REGION \
--project $PROJECT_ID
Expand All @@ -204,19 +204,19 @@ steps:
-var=jupyter_add_auth=false \
-var=frontend_add_auth=false \
-var=project_id=$PROJECT_ID \
-var=cluster_name=ml-$SHORT_SHA-$_PR_NUMBER-$_SUFFIX-cluster \
-var=kubernetes_namespace=rag-$SHORT_SHA-$_SUFFIX \
-var=gcs_bucket=gke-aieco-rag-$SHORT_SHA-$_SUFFIX \
-var=ray_service_account=ray-sa-$SHORT_SHA-$_SUFFIX \
-var=rag_service_account=rag-sa-$SHORT_SHA-$_SUFFIX \
-var=jupyter_service_account=jupyter-sa-$SHORT_SHA-$_SUFFIX \
-var=cloudsql_instance=pgvector-instance-$SHORT_SHA-$_SUFFIX \
-var=cluster_name=ml-$SHORT_SHA-$_PR_NUMBER-$BUILD_ID-cluster \
-var=kubernetes_namespace=rag-$SHORT_SHA-$BUILD_ID \
-var=gcs_bucket=gke-aieco-rag-$SHORT_SHA-$BUILD_ID \
-var=ray_service_account=ray-sa-$SHORT_SHA-$BUILD_ID \
-var=rag_service_account=rag-sa-$SHORT_SHA-$BUILD_ID \
-var=jupyter_service_account=jupyter-sa-$SHORT_SHA-$BUILD_ID \
-var=cloudsql_instance=pgvector-instance-$SHORT_SHA-$BUILD_ID \
-auto-approve -no-color -lock=false
echo "pass" > /workspace/rag_tf_result.txt
# Validate Ray: Make sure pods are running
kubectl wait --all pods -n rag-$SHORT_SHA-$_SUFFIX --for=condition=Ready --timeout=300s
kubectl port-forward -n rag-$SHORT_SHA-$_SUFFIX service/example-cluster-kuberay-head-svc 8265:8265 &
kubectl wait --all pods -n rag-$SHORT_SHA-$BUILD_ID --for=condition=Ready --timeout=300s
kubectl port-forward -n rag-$SHORT_SHA-$BUILD_ID service/example-cluster-kuberay-head-svc 8265:8265 &
# Wait port-forwarding to take its place
sleep 5s
Expand All @@ -226,8 +226,8 @@ steps:
echo "pass" > /workspace/rag_ray_dashboard_result.txt
# Validate Jupyterhub: Get hub url
kubectl get services -n rag-$SHORT_SHA-$_SUFFIX
kubectl get service proxy-public -n rag-$SHORT_SHA-$_SUFFIX --output jsonpath='{.status.loadBalancer.ingress[0].ip}' > /workspace/rag_jupyterhub_host_url.txt
kubectl get services -n rag-$SHORT_SHA-$BUILD_ID
kubectl get service proxy-public -n rag-$SHORT_SHA-$BUILD_ID --output jsonpath='{.status.loadBalancer.ingress[0].ip}' > /workspace/rag_jupyterhub_host_url.txt
echo "HOST URL is " $(cat /workspace/rag_jupyterhub_host_url.txt)
# Validate Jupyterhub: Test Hub
Expand All @@ -236,7 +236,7 @@ steps:
echo "pass" > /workspace/rag_jupyterhub_test_result.txt
# Validate RAG: Test rag frontend
kubectl port-forward -n rag-$SHORT_SHA-$_SUFFIX service/rag-frontend 8081:8080 &
kubectl port-forward -n rag-$SHORT_SHA-$BUILD_ID service/rag-frontend 8081:8080 &
# Wait port-forwarding to take its place
sleep 5s
Expand All @@ -258,13 +258,13 @@ steps:
terraform destroy \
-var-file=workloads.tfvars \
-var=project_id=$PROJECT_ID \
-var=cluster_name=ml-$SHORT_SHA-$_PR_NUMBER-$_SUFFIX-cluster \
-var=kubernetes_namespace=rag-$SHORT_SHA-$_SUFFIX \
-var=gcs_bucket=gke-aieco-rag-$SHORT_SHA-$_SUFFIX \
-var=ray_service_account=ray-sa-$SHORT_SHA-$_SUFFIX \
-var=rag_service_account=rag-sa-$SHORT_SHA-$_SUFFIX \
-var=jupyter_service_account=jupyter-sa-$SHORT_SHA-$_SUFFIX \
-var=cloudsql_instance=pgvector-instance-$SHORT_SHA-$_SUFFIX \
-var=cluster_name=ml-$SHORT_SHA-$_PR_NUMBER-$BUILD_ID-cluster \
-var=kubernetes_namespace=rag-$SHORT_SHA-$BUILD_ID \
-var=gcs_bucket=gke-aieco-rag-$SHORT_SHA-$BUILD_ID \
-var=ray_service_account=ray-sa-$SHORT_SHA-$BUILD_ID \
-var=rag_service_account=rag-sa-$SHORT_SHA-$BUILD_ID \
-var=jupyter_service_account=jupyter-sa-$SHORT_SHA-$BUILD_ID \
-var=cloudsql_instance=pgvector-instance-$SHORT_SHA-$BUILD_ID \
-auto-approve -no-color
allowFailure: true
Expand All @@ -280,7 +280,7 @@ steps:
cd /workspace/infrastructure
terraform destroy -var-file=tfvars_tests/standard-gke-public.platform.tfvars -var=project_id=$PROJECT_ID \
-var=cluster_name=ml-$SHORT_SHA-$_PR_NUMBER-$_SUFFIX-cluster \
-var=cluster_name=ml-$SHORT_SHA-$_PR_NUMBER-$BUILD_ID-cluster \
-var=autopilot_cluster=$_AUTOPILOT_CLUSTER \
-var=cluster_location=$_REGION -auto-approve -no-color
Expand Down Expand Up @@ -344,8 +344,7 @@ substitutions:
_REGION: us-central1
_USER_NAME: github
_AUTOPILOT_CLUSTER: "false"
_SUFFIX: std
options:
substitutionOption: 'ALLOW_LOOSE'
machineType: 'E2_HIGHCPU_8'
timeout: 5400s
timeout: 5400s
1 change: 0 additions & 1 deletion modules/jupyter/tests/change_jupyter_config.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -7,7 +7,6 @@
if autopilot:
config_file = "../jupyter_config/config-selfauth-autopilot.yaml"


with open(config_file, "r") as yaml_file:
data = yaml.safe_load(yaml_file)

Expand Down

0 comments on commit 4a6fe4b

Please sign in to comment.