feat: add Docker Compose deployment, refactor config paths and mock modes to CLI

.dockerignore

@@ -0,0 +1,25 @@
+**/.DS_Store
+**/.classpath
+**/.dockerignore
+**/.env
+**/.git
+**/.gitignore
+**/.project
+**/.settings
+**/.toolstarget
+**/.vs
+**/.vscode
+**/*.*proj.user
+**/*.dbmdl
+**/*.jfm
+**/charts
+**/docker-compose*
+**/compose.y*ml
+**/Dockerfile*
+**/node_modules
+**/npm-debug.log
+**/secrets.dev.yaml
+**/values.dev.yaml
+/bin
+/target
+LICENSE

.env.example

@@ -0,0 +1,50 @@
+# DockerHub image settings
+DOCKERHUB_USERNAME=hbwei
+IMAGE_TAG=v0.0.1
+
+# Host user/group IDs — set these to match your host user (id -u / id -g)
+# so that bind-mounted files are accessible inside containers.
+HOST_UID=1000
+HOST_GID=1000
+
+# quic client-side only
+# Same-host: use Docker bridge gateway (10.64.89.1). Remote: use server host's routable IP.
+QUIC_CLIENT_REMOTE_ADDR=10.64.89.1:12345
+
+QUIC_DOCKER_CLIENT_CONFIG_PATH=./docker/config/quic_config_client_docker.yaml
+
+# quic server-side only
+QUIC_SERVER_ADDR=0.0.0.0:12345
+QUIC_SERVER_PORT=12345
+
+QUIC_DOCKER_SERVER_CONFIG_PATH=./docker/config/quic_config_gcloud_docker.yaml
+
+# python client config
+PYTHON_DOCKER_CLIENT_CONFIG_PATH=./docker/config/client_config_docker.yaml
+
+# python server config
+PYTHON_DOCKER_SERVER_CONFIG_PATH=./docker/config/server_config_gcloud_docker.yaml
+
+# both client and server-side
+DOCKER_LOG_CONFIG_PATH=./docker/config/logging_config_docker.yaml
+EXPERIMENT_OUTPUT_DIR=/path/to/experiment2-out
+
+# client-side only (required even in mock camera mode — bandwidth allocator needs eval data)
+DASHBOARD_PORT=5000
+MODEL_FULL_EVAL_DIR=/path/to/full-eval
+
+# server-side only (not needed in mock inference mode, but must point to a valid directory — an empty one is fine)
+EFFDET_MODELS_DIR=/path/to/av-models
+
+# rust env vars
+RUST_LOG=info
+RUST_BACKTRACE=full
+
+# for both client and server
+EXPERIMENT_MODEL_INFO_PATH=./experiment_model_info.csv
+MOCK_EFFDET_OUTPUT_PATH=./src/python/camera_stream/example_effdet_d4_output.npy
+MOCK_EFFDET_INPUT_PATH=./src/python/camera_stream/mock_webcam_image.jpg
+
+# Mock modes — set to any non-empty value (e.g. "true") to enable, leave empty to disable
+MOCK_CAMERA=
+MOCK_INFERENCE=

.gitignore

@@ -36,7 +36,6 @@ bandwidth_allocation/.DS_Store
 
 **/.vscode/*
 **/target/*
-**/Cargo.lock
 
 # we don't want your SSL certs/keys
 **/*.key

compose.gpu.yaml

@@ -0,0 +1,11 @@
+# GPU override — include this file on hosts with NVIDIA GPUs.
+# Usage: docker compose -f compose.yaml -f compose.gpu.yaml --profile server up
+services:
+  server_python_main:
+    deploy:
+      resources:
+        reservations:
+          devices:
+            - driver: nvidia
+              count: all
+              capabilities: [gpu]

compose.yaml

@@ -0,0 +1,179 @@
+services:
+  quic_client:
+    profiles: ['client']
+    init: true
+    ipc: host
+    stop_grace_period: 30s
+    user: "${HOST_UID}:${HOST_GID}"
+    image: ${DOCKERHUB_USERNAME}/turbo-quic-client:${IMAGE_TAG:-latest}
+    depends_on:
+      client_python_monitor:
+        condition: service_healthy
+    environment:
+      RUST_LOG: ${RUST_LOG}
+      RUST_BACKTRACE: ${RUST_BACKTRACE}
+      QUIC_BIND_ADDR: ${QUIC_CLIENT_REMOTE_ADDR}
+    volumes:
+      - type: bind
+        source: ${QUIC_DOCKER_CLIENT_CONFIG_PATH}
+        target: /app/quic_config.yaml
+        read_only: true
+      - type: bind
+        source: ${EXPERIMENT_OUTPUT_DIR}
+        target: /app/experiment2-out
+      - zmq_client:/app/zmq-out
+      - health_signals:/health
+    networks:
+      - quic_net
+
+  quic_server:
+    profiles: ['server']
+    init: true
+    ipc: host
+    stop_grace_period: 30s
+    user: "${HOST_UID}:${HOST_GID}"
+    image: ${DOCKERHUB_USERNAME}/turbo-quic-server:${IMAGE_TAG:-latest}
+    environment:
+      RUST_LOG: ${RUST_LOG}
+      RUST_BACKTRACE: ${RUST_BACKTRACE}
+    volumes:
+      - type: bind
+        source: ${QUIC_DOCKER_SERVER_CONFIG_PATH}
+        target: /app/quic_config.yaml
+        read_only: true
+      - type: bind
+        source: ${EXPERIMENT_OUTPUT_DIR}
+        target: /app/experiment2-out
+      - zmq_server:/app/zmq-out
+    ports:
+      - "${QUIC_SERVER_PORT}:${QUIC_SERVER_PORT}/udp"
+
+  client_python_main:
+    profiles: ['client']
+    init: true
+    ipc: host
+    stop_grace_period: 30s
+    user: "${HOST_UID}:${HOST_GID}"
+    image: ${DOCKERHUB_USERNAME}/turbo-python-client:${IMAGE_TAG:-latest}
+    volumes:
+      - health_signals:/health
+      - type: bind
+        source: ${PYTHON_DOCKER_CLIENT_CONFIG_PATH}
+        target: /app/python_config.yaml
+        read_only: true
+      - type: bind
+        source: ${DOCKER_LOG_CONFIG_PATH}
+        target: /app/logging_config.yaml
+        read_only: true
+      - type: bind
+        source: ${EXPERIMENT_OUTPUT_DIR}
+        target: /app/experiment2-out
+      - zmq_client:/app/zmq-out
+      - type: bind
+        source: ${MODEL_FULL_EVAL_DIR}
+        target: /app/full-eval
+        read_only: true
+      - type: bind
+        source: ${EXPERIMENT_MODEL_INFO_PATH}
+        target: /app/experiment_model_info.csv
+        read_only: true
+      - type: bind
+        source: ${MOCK_EFFDET_INPUT_PATH}
+        target: /app/mock_webcam_image.jpg
+        read_only: true
+    command: >
+      bash -c 'cd src/python && exec python client_main.py -c /app/python_config.yaml -s ${QUIC_CLIENT_REMOTE_ADDR} ${MOCK_CAMERA:+--mock-camera}'
+    healthcheck:
+      test: ["CMD", "test", "-f", "/health/client_main_ready"]
+      interval: 2s
+      timeout: 3s
+      start_period: 30s
+      retries: 30
+
+  server_python_main:
+    profiles: ['server']
+    init: true
+    ipc: host
+    stop_grace_period: 30s
+    user: "${HOST_UID}:${HOST_GID}"
+    image: ${DOCKERHUB_USERNAME}/turbo-python-server:${IMAGE_TAG:-latest}
+    volumes:
+      - type: bind
+        source: ${PYTHON_DOCKER_SERVER_CONFIG_PATH}
+        target: /app/python_config.yaml
+        read_only: true
+      - type: bind
+        source: ${DOCKER_LOG_CONFIG_PATH}
+        target: /app/logging_config.yaml
+        read_only: true
+      - type: bind
+        source: ${EXPERIMENT_OUTPUT_DIR}
+        target: /app/experiment2-out
+      - zmq_server:/app/zmq-out
+      - type: bind
+        source: ${EFFDET_MODELS_DIR}
+        target: /app/av-models
+        read_only: true
+      - type: bind
+        source: ${EXPERIMENT_MODEL_INFO_PATH}
+        target: /app/experiment_model_info.csv
+        read_only: true
+      - type: bind
+        source: ${MOCK_EFFDET_OUTPUT_PATH}
+        target: /app/example_effdet_d4_output.npy
+        read_only: true
+    command: >
+      bash -c 'cd src/python && exec python server_main.py -c /app/python_config.yaml ${MOCK_INFERENCE:+--mock-inference}'
+
+  client_python_monitor:
+    profiles: ['client']
+    init: true
+    ipc: host
+    stop_grace_period: 30s
+    user: "${HOST_UID}:${HOST_GID}"
+    image: ${DOCKERHUB_USERNAME}/turbo-python-monitor:${IMAGE_TAG:-latest}
+    depends_on:
+      client_python_main:
+        condition: service_healthy
+    volumes:
+      - health_signals:/health
+      - type: bind
+        source: ${PYTHON_DOCKER_CLIENT_CONFIG_PATH}
+        target: /app/python_config.yaml
+        read_only: true
+      - type: bind
+        source: ${EXPERIMENT_OUTPUT_DIR}
+        target: /app/experiment2-out
+      - zmq_client:/app/zmq-out
+    ports:
+      - "${DASHBOARD_PORT}:5000"
+    healthcheck:
+      test: ["CMD", "test", "-f", "/health/monitor_ready"]
+      interval: 2s
+      timeout: 3s
+      start_period: 30s
+      retries: 30
+
+networks:
+  quic_net:
+    ipam:
+      config:
+        - subnet: 10.64.89.0/24
+          gateway: 10.64.89.1
+
+volumes:
+  health_signals:
+    driver: local
+    driver_opts:
+      type: tmpfs
+      device: tmpfs
+  zmq_client:
+    driver: local
+    driver_opts:
+      type: tmpfs
+      device: tmpfs
+  zmq_server:
+    driver: local
+    driver_opts:
+      type: tmpfs
+      device: tmpfs

config/client_config.yaml

@@ -19,9 +19,6 @@ MAX_LOG_ENTRIES: &max-log-entries 100
 SLO_TIMEOUT: &slo-timeout 200 # TODO: change this
 QUIC_SHM_SIZE: &quic-shm-size 50000000
 
-# needed for ping loop
-DST_IP: &dst-ip 136.118.43.70
-
 main_client_config_list:
   - service_id: 1
     max_entries: *max-log-entries
@@ -87,7 +84,6 @@ bandwidth_allocator_config:
   bidirectional_zmq_ping_handler_sockname: ping-handler
 
 ping_handler_config:
-  dst_ip: *dst-ip
   max_entries: 100
   thread_concurrency: 5
   bidirectional_zmq_sockname: ping-handler
@@ -103,7 +99,7 @@ camera_stream_config_list:
     shmem_buf_size: *quic-shm-size
     camera_np_size: [1080, 1920, 3]
     zmq_kill_switch_sockname: camera-kill-1-switch
-    mock_camera_image_path: null # set to a file path (e.g. mock_webcam_image.jpg) to use a static image instead of webcam
+    mock_camera_image_path: /home/hwei/turbo/src/python/camera_stream/mock_webcam_image.jpg # only used when --mock-camera is passed
   - camera_id: 2 # FRONT LEFT
     usb_id: 4 # TODO: CONFIGURE THIS
     max_entries: *max-log-entries
@@ -113,7 +109,7 @@ camera_stream_config_list:
     shmem_buf_size: *quic-shm-size
     camera_np_size: [1080, 1920, 3]
     zmq_kill_switch_sockname: camera-kill-2-switch
-    mock_camera_image_path: null
+    mock_camera_image_path: /home/hwei/turbo/src/python/camera_stream/mock_webcam_image.jpg # only used when --mock-camera is passed
   - camera_id: 3 # FRONT RIGHT
     usb_id: 8 # TODO: CONFIGURE THIS
     max_entries: *max-log-entries
@@ -123,7 +119,7 @@ camera_stream_config_list:
     shmem_buf_size: *quic-shm-size
     camera_np_size: [1080, 1920, 3]
     zmq_kill_switch_sockname: camera-kill-3-switch
-    mock_camera_image_path: null
+    mock_camera_image_path: /home/hwei/turbo/src/python/camera_stream/mock_webcam_image.jpg # only used when --mock-camera is passed
 
 main_plotter_config:
 # service 4 is for junk stream

config/server_config_gcloud.yaml

@@ -38,8 +38,8 @@ server_config_list:
     thread_concurrency: 10
     shm_filesize: *shm-filesize
     zmq_kill_switch_sockname: remote-server-kill-switch-1
-    mock_inference_output_path: null # set to a file path (e.g. example_effdet_d4_output.npz) to skip model loading and return mock detections
-    mock_model_latency_csv_path: null # set to experiment_model_info.csv path to simulate per-model inference latency in mock mode
+    mock_inference_output_path: /home/hwei/turbo/src/python/camera_stream/example_effdet_d4_output.npy # only used when --mock-inference is passed
+    mock_model_latency_csv_path: /home/hwei/turbo/experiment_model_info.csv # only used when --mock-inference is passed
   - service_id: 2
     max_entries: *max-log-entries
     model_metadata_list: *model-list-ref
@@ -51,8 +51,8 @@ server_config_list:
     thread_concurrency: 10
     shm_filesize: *shm-filesize
     zmq_kill_switch_sockname: remote-server-kill-switch-2
-    mock_inference_output_path: null
-    mock_model_latency_csv_path: null
+    mock_inference_output_path: /home/hwei/turbo/src/python/camera_stream/example_effdet_d4_output.npy # only used when --mock-inference is passed
+    mock_model_latency_csv_path: /home/hwei/turbo/experiment_model_info.csv # only used when --mock-inference is passed
   - service_id: 3
     max_entries: *max-log-entries
     model_metadata_list: *model-list-ref
@@ -64,5 +64,5 @@ server_config_list:
     thread_concurrency: 10
     shm_filesize: *shm-filesize
     zmq_kill_switch_sockname: remote-server-kill-switch-3
-    mock_inference_output_path: null
-    mock_model_latency_csv_path: null
+    mock_inference_output_path: /home/hwei/turbo/src/python/camera_stream/example_effdet_d4_output.npy # only used when --mock-inference is passed
+    mock_model_latency_csv_path: /home/hwei/turbo/experiment_model_info.csv # only used when --mock-inference is passed