Add script to verify async indexing configured

ann_benchmark.sh

@@ -24,10 +24,19 @@ echo "Building all required containers"
 ( cd apps/ann-benchmarks/ && docker build -t ann_benchmarks . )
 
 echo "Starting Weaviate..."
+echo "ASYNC_INDEXING: $ASYNC_INDEXING"
 docker compose -f apps/weaviate-no-restart-on-crash/docker-compose.yml up -d
 
 wait_weaviate
 
+echo "Current directory: $PWD"
+echo "Exising files in current directory: $(ls -la)"
+
+echo "Printing environment variables"
+apt install python3-pip -y
+python3 -m pip install docker
+python3 get_env_var.py
+
 echo "Run benchmark script"
 mkdir -p datasets
 ( 

ann_benchmark_gcp.sh

@@ -43,6 +43,7 @@ gcloud compute ssh --zone $ZONE $instance -- "mkdir -p ~/apps/"
 gcloud compute scp --zone $ZONE --recurse apps/ann-benchmarks "$instance:~/apps/"
 gcloud compute scp --zone $ZONE --recurse apps/weaviate-no-restart-on-crash/ "$instance:~/apps/"
 gcloud compute scp --zone $ZONE --recurse ann_benchmark.sh "$instance:~"
-gcloud compute ssh --zone $ZONE $instance -- "DATASET=$DATASET DISTANCE=$DISTANCE REQUIRED_RECALL=$REQUIRED_RECALL WEAVIATE_VERSION=$WEAVIATE_VERSION MACHINE_TYPE=$MACHINE_TYPE CLOUD_PROVIDER=$CLOUD_PROVIDER OS=$OS bash ann_benchmark.sh"
+gcloud compute scp --zone $ZONE --recurse get_env_var.py "$instance:~"
+gcloud compute ssh --zone $ZONE $instance -- "ASYNC_INDEXING=$ASYNC_INDEXING DATASET=$DATASET DISTANCE=$DISTANCE REQUIRED_RECALL=$REQUIRED_RECALL WEAVIATE_VERSION=$WEAVIATE_VERSION MACHINE_TYPE=$MACHINE_TYPE CLOUD_PROVIDER=$CLOUD_PROVIDER OS=$OS bash ann_benchmark.sh"
 mkdir -p results
 gcloud compute scp --zone $ZONE --recurse "$instance:~/results/*.json" results/

apps/ann-benchmarks/requirements.txt

@@ -3,3 +3,4 @@ loguru==0.5.3
 seaborn==0.12.2
 h5py==3.11.0
 pandas==2.2.2
+docker==7.1.0

apps/get_env_var.sh

@@ -0,0 +1,25 @@
+import docker
+
+def get_env_variables(partial_image_name):
+    client = docker.from_env()
+    # List all running containers
+    containers = client.containers.list()
+
+    # Filter containers where the image name matches the partial name
+    matching_containers = [
+        container for container in containers if partial_image_name in container.image.tags[0]
+    ]
+
+    if not matching_containers:
+        print(f"No running containers found matching image name: {partial_image_name}")
+        return
+
+    # Retrieve and print environment variables for each matching container
+    for container in matching_containers:
+        print(f"Environment variables for container {container.short_id}:")
+        exec_result = container.exec_run("env", stdout=True, stderr=True)
+        print(exec_result.output.decode())
+        print("--------------------------------------------")
+
+# Replace with a partial image name
+get_env_variables("weaviate")

get_env_var.py

@@ -0,0 +1,22 @@
+import docker
+
+def get_env_variables(partial_image_name):
+    client = docker.from_env()
+    containers = client.containers.list()
+
+    matching_containers = [
+        container for container in containers if partial_image_name in container.image.tags[0]
+    ]
+
+    if not matching_containers:
+        print(f"No running containers found matching image name: {partial_image_name}")
+        return
+
+    for container in matching_containers:
+        print(f"Environment variables for container {container.short_id}:")
+        exec_result = container.exec_run("env", stdout=True, stderr=True)
+        print(exec_result.output.decode())
+        print("--------------------------------------------")
+
+get_env_variables("weaviate")
+