Enhance CI Workflow for Scenic Simulators: Improved Volume Management, Reliability, and Cost Efficiency

.github/workflows/run-simulators.yml

@@ -10,13 +10,42 @@ jobs:
       runs-on: ubuntu-latest
       concurrency:
         group: sim
+      outputs:
+        volume_id: ${{ steps.create_volume_step.outputs.volume_id }}
+      env:
+        INSTANCE_ID: ${{ secrets.AWS_EC2_INSTANCE_ID }}
+        AWS_ACCESS_KEY_ID: ${{ secrets.AWS_ACCESS_KEY_ID }}
+        AWS_SECRET_ACCESS_KEY: ${{ secrets.AWS_SECRET_ACCESS_KEY }}
+        AWS_DEFAULT_REGION: ${{ secrets.AWS_REGION }}
       steps:
+        - name: Create Volume from Latest Snapshot and Attach to Instance
+          id: create_volume_step
+          run: |
+            #  Retrieve the latest snapshot ID
+            LATEST_SNAPSHOT_ID=$(aws ec2 describe-snapshots --owner-ids self --query 'Snapshots | sort_by(@, &StartTime) | [-1].SnapshotId' --output text)
+            echo "Checking availability for snapshot: $LATEST_SNAPSHOT_ID"
+
+            # Wait for the snapshot to complete
+            aws ec2 wait snapshot-completed --snapshot-ids $LATEST_SNAPSHOT_ID
+            echo "Snapshot is ready."
+
+            # Create a new volume from the latest snapshot
+            volume_id=$(aws ec2 create-volume --snapshot-id $LATEST_SNAPSHOT_ID --availability-zone us-west-1b --volume-type gp3 --size 400 --throughput 250 --query "VolumeId" --output text)
+            echo "Created volume with ID: $volume_id"
+
+            # Set volume_id as output
+            echo "volume_id=$volume_id" >> $GITHUB_OUTPUT
+            cat $GITHUB_OUTPUT
+
+            # Wait until the volume is available
+            aws ec2 wait volume-available --volume-ids $volume_id
+            echo "Volume is now available"
+
+            # Attach the volume to the instance
+            aws ec2 attach-volume --volume-id $volume_id --instance-id $INSTANCE_ID --device /dev/sda1
+            echo "Volume $volume_id attached to instance $INSTANCE_ID as /dev/sda1"
+
         - name: Start EC2 Instance
-          env:
-            INSTANCE_ID: ${{ secrets.AWS_EC2_INSTANCE_ID }}
-            AWS_ACCESS_KEY_ID: ${{ secrets.AWS_ACCESS_KEY_ID }}
-            AWS_SECRET_ACCESS_KEY: ${{ secrets.AWS_SECRET_ACCESS_KEY }}
-            AWS_DEFAULT_REGION: ${{ secrets.AWS_REGION }}
           run: |
             # Get the instance state
             instance_state=$(aws ec2 describe-instances --instance-ids $INSTANCE_ID | jq -r '.Reservations[].Instances[].State.Name')
@@ -27,7 +56,7 @@ jobs:
                 sleep 10
                 instance_state=$(aws ec2 describe-instances --instance-ids $INSTANCE_ID | jq -r '.Reservations[].Instances[].State.Name')
             done
-            
+
             # Check if instance state is "stopped"
             if [[ "$instance_state" == "stopped" ]]; then
                 echo "Instance is stopped, starting it..."
@@ -42,34 +71,17 @@ jobs:
                 exit 1
             fi
 
-            # wait for status checks to pass 
-            TIMEOUT=300  # Timeout in seconds
-            START_TIME=$(date +%s)
-            END_TIME=$((START_TIME + TIMEOUT))
-            while true; do
-                response=$(aws ec2 describe-instance-status --instance-ids $INSTANCE_ID)
-                system_status=$(echo "$response" | jq -r '.InstanceStatuses[0].SystemStatus.Status')
-                instance_status=$(echo "$response" | jq -r '.InstanceStatuses[0].InstanceStatus.Status')
-
-                if [[ "$system_status" == "ok" && "$instance_status" == "ok" ]]; then
-                    echo "Both SystemStatus and InstanceStatus are 'ok'"
-                    exit 0
-                fi
-
-                CURRENT_TIME=$(date +%s)
-                if [[ "$CURRENT_TIME" -ge "$END_TIME" ]]; then
-                    echo "Timeout: Both SystemStatus and InstanceStatus have not reached 'ok' state within $TIMEOUT seconds."
-                    exit 1
-                fi
-
-                sleep 10  # Check status every 10 seconds
-            done
+            # Wait for instance status checks to pass
+            echo "Waiting for instance status checks to pass..."
+            aws ec2 wait instance-status-ok --instance-ids $INSTANCE_ID
+            echo "Instance is now ready for use."
+
 
     check_simulator_version_updates:
       name: check_simulator_version_updates
       runs-on: ubuntu-latest
       needs: start_ec2_instance
-      steps: 
+      steps:
         - name: Check for Simulator Version Updates
           env:
             PRIVATE_KEY: ${{ secrets.SSH_PRIVATE_KEY }}
@@ -109,11 +121,11 @@ jobs:
                 echo "NVIDIA Driver is not set"
                 exit 1
               fi
-            '  
+            '
         - name: NVIDIA Driver is not set
           if: ${{ failure() }}
           run: |
-            echo "NVIDIA SMI is not working, please run the steps here on the instance:" 
+            echo "NVIDIA SMI is not working, please run the steps here on the instance:"
             echo "https://scenic-lang.atlassian.net/wiki/spaces/KAN/pages/2785287/Setting+Up+AWS+VM?parentProduct=JSW&initialAllowedFeatures=byline-contributors.byline-extensions.page-comments.delete.page-reactions.inline-comments.non-licensed-share&themeState=dark%253Adark%2520light%253Alight%2520spacing%253Aspacing%2520colorMode%253Alight&locale=en-US#Install-NVIDIA-Drivers"
 
     run_carla_simulators:
@@ -128,17 +140,17 @@ jobs:
             USER_NAME: ${{secrets.SSH_USERNAME}}
           run: |
             echo "$PRIVATE_KEY" > private_key && chmod 600 private_key
-            ssh -o StrictHostKeyChecking=no -i private_key ${USER_NAME}@${HOSTNAME} '
+            ssh -o StrictHostKeyChecking=no -o ServerAliveInterval=60 -o ServerAliveCountMax=3 -i private_key ${USER_NAME}@${HOSTNAME} '
               cd /home/ubuntu/actions/Scenic &&
               source venv/bin/activate &&
               carla_versions=($(find /software -maxdepth 1 -type d -name 'carla*')) &&
               for version in "${carla_versions[@]}"; do
-                echo "============================= CARLA $version ============================="  
+                echo "============================= CARLA $version ============================="
                 export CARLA_ROOT="$version"
                 pytest tests/simulators/carla
               done
             '
-            
+
     run_webots_simulators:
       name: run_webots_simulators
       runs-on: ubuntu-latest
@@ -164,39 +176,44 @@ jobs:
               done
               kill %1
             '
-      
+
     stop_ec2_instance:
       name: stop_ec2_instance
       runs-on: ubuntu-latest
-      needs: [run_carla_simulators, run_webots_simulators]
-      steps:       
+      needs: [start_ec2_instance, check_simulator_version_updates, check_nvidia_smi, run_carla_simulators, run_webots_simulators]
+      if: always()
+      env:
+        VOLUME_ID: ${{ needs.start_ec2_instance.outputs.volume_id }}
+        INSTANCE_ID: ${{ secrets.AWS_EC2_INSTANCE_ID }}
+        AWS_ACCESS_KEY_ID: ${{ secrets.AWS_ACCESS_KEY_ID }}
+        AWS_SECRET_ACCESS_KEY: ${{ secrets.AWS_SECRET_ACCESS_KEY }}
+        AWS_DEFAULT_REGION: ${{ secrets.AWS_REGION }}
+      steps:
         - name: Stop EC2 Instance
-          env:
-            INSTANCE_ID: ${{ secrets.AWS_EC2_INSTANCE_ID }}
-            AWS_ACCESS_KEY_ID: ${{ secrets.AWS_ACCESS_KEY_ID }}
-            AWS_SECRET_ACCESS_KEY: ${{ secrets.AWS_SECRET_ACCESS_KEY }}
-            AWS_DEFAULT_REGION: ${{ secrets.AWS_REGION }}
           run: |
-            # Get the instance state
+            # Get the instance state and stop it if running
             instance_state=$(aws ec2 describe-instances --instance-ids $INSTANCE_ID | jq -r '.Reservations[].Instances[].State.Name')
-
-            # If the machine is pending wait for it to fully start
-            while [ "$instance_state" == "pending" ]; do
-                echo "Instance is pending startup, waiting for it to fully start..."
-                sleep 10
-                instance_state=$(aws ec2 describe-instances --instance-ids $INSTANCE_ID | jq -r '.Reservations[].Instances[].State.Name')
-            done
-
-            # Check if instance state is "stopped"
             if [[ "$instance_state" == "running" ]]; then
-                echo "Instance is running, stopping it..."
-                aws ec2 stop-instances --instance-ids $INSTANCE_ID
-            elif [[ "$instance_state" == "stopping" ]]; then
-                echo "Instance is stopping..."
+              echo "Instance is running, stopping it..."
+              aws ec2 stop-instances --instance-ids $INSTANCE_ID
+              aws ec2 wait instance-stopped --instance-ids $INSTANCE_ID
+              echo "Instance has stopped."
             elif [[ "$instance_state" == "stopped" ]]; then
-                echo "Instance is already stopped..."
-                exit 0
+              echo "Instance is already stopped."
             else
-                echo "Unknown instance state: $instance_state"
-                exit 1
+              echo "Unexpected instance state: $instance_state"
+              exit 1
             fi
+
+        - name: Detach Volume
+          run: |
+            # Detach the volume
+            aws ec2 detach-volume --volume-id $VOLUME_ID
+            aws ec2 wait volume-available --volume-ids $VOLUME_ID
+            echo "Volume $VOLUME_ID detached."
+
+        - name: Delete Volume
+          run: |
+            # Delete the volume after snapshot is complete
+            aws ec2 delete-volume --volume-id $VOLUME_ID
+            echo "Volume $VOLUME_ID deleted."

tests/simulators/carla/test_actions.py

@@ -43,19 +43,21 @@ def getCarlaSimulator(getAssetPath):
             f"bash {CARLA_ROOT}/CarlaUE4.sh -RenderOffScreen", shell=True
         )
 
-        for _ in range(30):
+        for _ in range(180):
             if isCarlaServerRunning():
                 break
             time.sleep(1)
+        else:
+            pytest.fail("Unable to connect to CARLA.")
 
         # Extra 5 seconds to ensure server startup
-        time.sleep(5)
+        time.sleep(10)
 
     base = getAssetPath("maps/CARLA")
 
     def _getCarlaSimulator(town):
         path = os.path.join(base, f"{town}.xodr")
-        simulator = CarlaSimulator(map_path=path, carla_map=town)
+        simulator = CarlaSimulator(map_path=path, carla_map=town, timeout=180)
         return simulator, town, path
 
     yield _getCarlaSimulator
@@ -76,7 +78,7 @@ def test_throttle(getCarlaSimulator):
         behavior DriveWithThrottle():
             while True:
                 take SetThrottleAction(1)
-        
+
         ego = new Car at (369, -326), with behavior DriveWithThrottle
         record ego.speed as CarSpeed
         terminate after 5 steps
@@ -109,8 +111,8 @@ def test_brake(getCarlaSimulator):
             do DriveWithThrottle() for 2 steps
             do Brake() for 6 steps
 
-        ego = new Car at (369, -326), 
-            with blueprint 'vehicle.toyota.prius', 
+        ego = new Car at (369, -326),
+            with blueprint 'vehicle.toyota.prius',
             with behavior DriveThenBrake
         record final ego.speed as CarSpeed
         terminate after 8 steps