feat: add docker CLI for building docker containers

.github/workflows/docker-image.yml

@@ -0,0 +1,46 @@
+name: Docker Image CI
+
+on:
+  push:
+    branches: ["main"]
+  pull_request:
+    branches: ["main"]
+
+jobs:
+  build:
+    runs-on: ubuntu-latest
+    steps:
+    - 
+      uses: actions/checkout@v4
+    -
+      name: Set up Docker Buildx
+      uses: docker/setup-buildx-action@v3
+    -
+      name: Login to Docker Hub
+      uses: docker/login-action@v3
+      with:
+        registry: ghcr.io
+        username: ${{ github.actor }}
+        password: ${{ secrets.GITHUB_TOKEN }}
+    -
+      name: Build prod and push
+      if: github.ref == 'refs/heads/main'
+      uses: docker/build-push-action@v6
+      with:
+        context: .
+        file: Dockerfile
+        tags: ghcr.io/ucsd-e4e/maestro_trainer:main
+        push: true
+        cache-from: type=registry,ref=ghcr.io/ucsd-e4e/maestro_trainer:main_cache
+        cache-to: type=registry,ref=ghcr.io/ucsd-e4e/maestro_trainer:main_cache,mode=max
+    -
+      name: Build prod and push
+      if: github.ref != 'refs/heads/main'
+      uses: docker/build-push-action@v6
+      with:
+        context: .
+        file: Dockerfile
+        tags: ghcr.io/ucsd-e4e/maestro_trainer:${{ github.head_ref }}
+        push: true
+        cache-from: type=registry,ref=ghcr.io/ucsd-e4e/maestro_trainer:${{ github.head_ref }}_cache
+        cache-to: type=registry,ref=ghcr.io/ucsd-e4e/maestro_trainer:${{ github.head_ref }}_cache,mode=max

model_trainer/Dockerfile → Dockerfile

@@ -11,6 +11,6 @@ WORKDIR /app
 ADD . /app
 RUN ls -la .
 
-ENTRYPOINT ["python", "src/model_trainer.py"] 
+ENTRYPOINT ["python", "model_trainer/src/model_trainer.py"] 
 # CMD ["python", "src/model_trainer.py"] 
 # 

README.md

@@ -0,0 +1,3 @@
+# model_trainer
+
+An example trainer for maestro active learning system

model_trainer/src/model_trainer.py

@@ -1,50 +1,50 @@
 
 import os
-import socketio
-import time 
-from model import Trainer
+# import socketio
+# import time 
+# from model import Trainer
 
 print(os.environ)
 print("RUNNING PYTHON!!!", flush=True)
 
 
 
 
-pod_name = os.environ['pod-name']
-sio = socketio.Client()
+# pod_name = os.environ['pod-name']
+# sio = socketio.Client()
 
-@sio.event
-def connect():
-    print('connection established', flush=True)
-    sio.emit('job_ready', {'pod-name': pod_name})
+# @sio.event
+# def connect():
+#     print('connection established', flush=True)
+#     sio.emit('job_ready', {'pod-name': pod_name})
 
-@sio.on('start_job')
-def trigger_test(data):
-    print("importing in model")
-    start = time.time()
-    print(data, flush=True)
-    # Expect batch to be a list of directory of file and label
-    batch = data["batch"]
-    cfg = data["cfg"]
-    trainer = Trainer(cfg, batch)
-    trainer.build()
-    loss = trainer.train()
+# @sio.on('start_job')
+# def trigger_test(data):
+#     print("importing in model")
+#     start = time.time()
+#     print(data, flush=True)
+#     # Expect batch to be a list of directory of file and label
+#     batch = data["batch"]
+#     cfg = data["cfg"]
+#     trainer = Trainer(cfg, batch)
+#     trainer.build()
+#     loss = trainer.train()
 
-    end = time.time()
-    print("Job Done", loss, "time", end - start, flush=True)
-    sio.emit('job_done', {'loss': loss, "pod-name": pod_name}, callback=sio.disconnect)
+#     end = time.time()
+#     print("Job Done", loss, "time", end - start, flush=True)
+#     sio.emit('job_done', {'loss': loss, "pod-name": pod_name}, callback=sio.disconnect)
 
-@sio.event
-def my_message(data):
-    print('message received with ', data, flush=True)
-    sio.emit('my response', {'response': 'my response'})
+# @sio.event
+# def my_message(data):
+#     print('message received with ', data, flush=True)
+#     sio.emit('my response', {'response': 'my response'})
 
-@sio.event
-def disconnect():
-    print('disconnected from server', flush=True)   
+# @sio.event
+# def disconnect():
+#     print('disconnected from server', flush=True)   
 
-sio.connect('https://test-service2.nrp-nautilus.io')
-sio.wait()
+# sio.connect('https://test-service2.nrp-nautilus.io')
+# sio.wait()