ci: dev docker containers tagged by hash and cleanup workflow

.github/workflows/cleanup-old-docker.yaml

@@ -0,0 +1,54 @@
+name: Cleanup Old Development Docker Images
+
+on:
+    workflow_dispatch:
+    schedule:
+        - cron: "0 2 * * *" # Run daily at 2 AM
+
+jobs:
+    cleanup:
+        if: ${{ github.repository == 'sgl-project/sglang' }}
+        runs-on: ubuntu-22.04
+        steps:
+            - name: Cleanup old dev images
+              env:
+                  DOCKERHUB_USERNAME: ${{ secrets.DOCKERHUB_USERNAME }}
+                  DOCKERHUB_TOKEN: ${{ secrets.DOCKERHUB_TOKEN }}
+              run: |
+                  # Get auth token
+                  TOKEN=$(curl -s -H "Content-Type: application/json" -X POST \
+                    -d '{"username": "'${DOCKERHUB_USERNAME}'", "password": "'${DOCKERHUB_TOKEN}'"}' \
+                    https://hub.docker.com/v2/users/login/ | jq -r .token)
+
+                  # Calculate cutoff date (2 weeks ago)
+                  CUTOFF_DATE=$(date -u -d '14 days ago' +%s)
+
+                  # Tags to clean up
+                  for TAG_PREFIX in "dev-x86" "dev-arm64" "dev"; do
+                    echo "Checking ${TAG_PREFIX}-* tags..."
+
+                    # Get all tags
+                    TAGS=$(curl -s -H "Authorization: JWT ${TOKEN}" \
+                      "https://hub.docker.com/v2/repositories/lmsysorg/sglang/tags/?page_size=100" \
+                      | jq -r '.results[] | select(.name | startswith("'${TAG_PREFIX}'-")) | select(.name != "'${TAG_PREFIX}'") | "\(.name)|\(.last_updated)"')
+
+                    # Process each tag
+                    echo "$TAGS" | while IFS='|' read -r TAG_NAME LAST_UPDATED; do
+                      if [ -z "$TAG_NAME" ]; then
+                        continue
+                      fi
+
+                      # Convert last_updated to epoch
+                      TAG_DATE=$(date -u -d "$LAST_UPDATED" +%s)
+
+                      # Check if older than 2 weeks
+                      if [ "$TAG_DATE" -lt "$CUTOFF_DATE" ]; then
+                        echo "Deleting old tag: $TAG_NAME (last updated: $LAST_UPDATED)"
+                        curl -X DELETE \
+                          -H "Authorization: JWT ${TOKEN}" \
+                          "https://hub.docker.com/v2/repositories/lmsysorg/sglang/tags/${TAG_NAME}/"
+                      else
+                        echo "Keeping tag: $TAG_NAME (last updated: $LAST_UPDATED)"
+                      fi
+                    done
+                  done

.github/workflows/release-docker-dev.yml

@@ -44,7 +44,14 @@ jobs:
 
       - name: Build and Push Dev Image (x86)
         run: |
-          docker buildx build --platform linux/amd64 --push -f docker/Dockerfile --build-arg CUDA_VERSION=${{ matrix.variant.version }} --build-arg BUILD_TYPE=${{ matrix.variant.type }} --build-arg CMAKE_BUILD_PARALLEL_LEVEL=$(nproc) -t lmsysorg/sglang:${{ matrix.variant.tag }} --no-cache .
+          COMMIT_SHA=$(git rev-parse --short HEAD)
+          docker buildx build --platform linux/amd64 --push -f docker/Dockerfile \
+            --build-arg CUDA_VERSION=${{ matrix.variant.version }} \
+            --build-arg BUILD_TYPE=${{ matrix.variant.type }} \
+            --build-arg CMAKE_BUILD_PARALLEL_LEVEL=$(nproc) \
+            -t lmsysorg/sglang:${{ matrix.variant.tag }} \
+            -t lmsysorg/sglang:${{ matrix.variant.tag }}-${COMMIT_SHA} \
+            --no-cache .
 
   build-dev-arm:
     if: ${{ github.repository == 'sgl-project/sglang' }}
@@ -84,7 +91,14 @@ jobs:
 
       - name: Build and Push Blackwell Image (ARM)
         run: |
-          docker buildx build --platform linux/arm64 --push -f docker/Dockerfile --build-arg CUDA_VERSION=${{ matrix.variant.version }} --build-arg BUILD_TYPE=${{ matrix.variant.type }} --build-arg CMAKE_BUILD_PARALLEL_LEVEL=$(nproc) -t lmsysorg/sglang:${{ matrix.variant.tag }} --no-cache .
+          COMMIT_SHA=$(git rev-parse --short HEAD)
+          docker buildx build --platform linux/arm64 --push -f docker/Dockerfile \
+            --build-arg CUDA_VERSION=${{ matrix.variant.version }} \
+            --build-arg BUILD_TYPE=${{ matrix.variant.type }} \
+            --build-arg CMAKE_BUILD_PARALLEL_LEVEL=$(nproc) \
+            -t lmsysorg/sglang:${{ matrix.variant.tag }} \
+            -t lmsysorg/sglang:${{ matrix.variant.tag }}-${COMMIT_SHA} \
+            --no-cache .
 
   create-manifests:
     runs-on: ubuntu-22.04
@@ -97,13 +111,24 @@ jobs:
             x86_tag: dev-x86
             arm64_tag: dev-arm64
     steps:
+      - name: Checkout repository
+        uses: actions/checkout@v4
+
       - uses: docker/setup-buildx-action@v3
       - uses: docker/login-action@v2
         with:
           username: ${{ secrets.DOCKERHUB_USERNAME }}
           password: ${{ secrets.DOCKERHUB_TOKEN }}
       - run: |
+          COMMIT_SHA=$(git rev-parse --short HEAD)
+          # Create regular dev manifest (latest)
           docker buildx imagetools create \
             -t lmsysorg/sglang:${{ matrix.variant.tag }} \
             lmsysorg/sglang:${{ matrix.variant.x86_tag }} \
             lmsysorg/sglang:${{ matrix.variant.arm64_tag }}
+
+          # Create commit-specific manifest
+          docker buildx imagetools create \
+            -t lmsysorg/sglang:${{ matrix.variant.tag }}-${COMMIT_SHA} \
+            lmsysorg/sglang:${{ matrix.variant.x86_tag }}-${COMMIT_SHA} \
+            lmsysorg/sglang:${{ matrix.variant.arm64_tag }}-${COMMIT_SHA}