feat(docker): build docker for dgx spark (#244)

Author: gufengc

.github/workflows/build-spark-image.yaml

@@ -0,0 +1,69 @@
+name: Build Spark Image
+
+on:
+  workflow_dispatch:
+  push:
+    branches:
+      - main
+    paths:
+      - 'docker/**'
+      - 'src/**'
+      - 'pyproject.toml'
+      - 'README.md'
+
+
+env:
+  IMAGE_NAME: parallax
+  NAMESPACE: gradientservice
+
+jobs:
+  build:
+    runs-on: arc-public-parallax
+    permissions:
+      contents: read
+      packages: write
+      id-token: write
+      attestations: write
+    strategy:
+      matrix:
+        variant: [spark]
+
+    steps:
+    - name: Checkout code
+      uses: actions/checkout@v4
+
+    - name: Set up Docker Buildx
+      uses: docker/setup-buildx-action@v3
+      with:
+        driver: remote
+        endpoint: tcp://buildkit-buildkit-service.arc-systems:1234
+
+
+    - name: Log in to Docker Hub
+      uses: docker/login-action@v3
+      with:
+        username: ${{ secrets.DOCKERHUB_USERNAME }}
+        password: ${{ secrets.DOCKERHUB_TOKEN }}
+
+    - name: Extract metadata
+      id: meta
+      uses: docker/metadata-action@v5
+      with:
+        images: ${{ env.NAMESPACE }}/${{ env.IMAGE_NAME }}
+        tags: |
+          type=ref,event=branch,suffix=-${{ matrix.variant }}
+          type=ref,event=pr,suffix=-${{ matrix.variant }}
+          type=raw,value=latest-${{ matrix.variant }},enable={{is_default_branch}}
+
+    - name: Build and push Docker image
+      id: build
+      uses: docker/build-push-action@v5
+      with:
+        context: .
+        file: ./docker/Dockerfile.${{ matrix.variant }}
+        push: true
+        tags: ${{ steps.meta.outputs.tags }}
+        labels: ${{ steps.meta.outputs.labels }}
+        cache-from: type=gha
+        cache-to: type=gha,mode=max
+        platforms: linux/arm64

docker/Dockerfile.spark

@@ -0,0 +1,23 @@
+FROM lmsysorg/sglang:spark
+
+ENV SGLANG_ENABLE_JIT_DEEPGEMM=0
+
+WORKDIR /parallax
+
+COPY README.md ./README.md
+COPY src ./src
+COPY pyproject.toml ./pyproject.toml
+
+RUN apt-get update
+
+RUN git clone --branch v0.29.1 https://github.com/ml-explore/mlx.git && \
+    apt-get install libblas-dev liblapack-dev liblapacke-dev -y && \
+    cd mlx && \
+    PYPI_RELEASE=1 pip install . --break-system-packages && \
+    cd ..
+
+# patch sglang version and model_parallel.py for current sglang spark docker
+RUN sed -i 's/sglang\[all\]==0.5.4.post1/sglang\[all\]==0.5.4.post2/g' pyproject.toml
+RUN sed -i 's/use_torch_symm_mem=use_torch_symm_mem/use_torch_symm_mem_all_reduce=use_torch_symm_mem/g' src/parallax/sglang/monkey_patch_utils/model_parallel.py
+
+RUN pip install -e '.[gpu]' --break-system-packages