add ability to push for e2e-gpu tests

.github/workflows/tests.yml

@@ -20,6 +20,37 @@ jobs:
       - name: Run integration tests
         run: make test-integration
 
+  e2e-gpu:
+    runs-on: ubuntu-latest
+    strategy:
+      matrix:
+        testcase:
+        - "engine-vllm-adapters"
+    # Timeout to address Github actions stuck scenarios.
+    timeout-minutes: 60
+    # Sets the permissions granted to the `GITHUB_TOKEN` for the actions in this job.
+    # This is required to push the image to the container registry.
+    permissions:
+      contents: read
+      packages: write
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@v4
+      - name: Set up Docker Buildx
+        uses: docker/setup-buildx-action@v3
+      - name: Log in to the Container registry
+        uses: docker/login-action@v3
+        with:
+          registry: ghcr.io
+          username: ${{ github.actor }}
+          password: ${{ secrets.GITHUB_TOKEN }}
+      - name: Run the e2e testcase
+        run: make test-e2e-${{ matrix.testcase }}
+        env:
+          TEST_KUBECONFIG: "${{ secrets.TEST_KUBECONFIG }}"
+          USE_GPU_CLUSTER: "true"
+          SKAFFOLD_DEFAULT_REPO: "ghcr.io"
+
   e2e-general:
     runs-on: ubuntu-latest
     # NOTE: Uncomment if we start getting limited on number of concurrent jobs
@@ -33,7 +64,6 @@ jobs:
         - "autoscaler-restart"
         - "cache-shared-filesystem"
         - "engine-vllm-pvc"
-        - "engine-vllm-adapters"
     steps:
       - name: Checkout code
         uses: actions/checkout@v4
@@ -55,8 +85,6 @@ jobs:
 
       - name: Run the e2e testcase
         run: make test-e2e-${{ matrix.testcase }}
-        env:
-          TEST_KUBECONFIG: ${{ secrets.TEST_KUBECONFIG }}
 
   e2e-engines:
     runs-on: ubuntu-latest

Makefile

@@ -96,6 +96,7 @@ test-e2e-engine-vllm-pvc: skaffold
 	./test/e2e/run.sh engine-vllm-pvc --profile e2e-test-default
 
 .PHONY: test-e2e-engine-vllm-adapters
+# The Lora adapters only work on vLLM with GPU.
 test-e2e-engine-vllm-adapters: export USE_GPU_CLUSTER=true
 test-e2e-engine-vllm-adapters: skaffold
 	./test/e2e/run.sh engine-vllm-adapters --profile e2e-test-gpu

skaffold.yaml

@@ -66,6 +66,9 @@ profiles:
         setValueTemplates:
           openwebui.enabled: false
 - name: e2e-test-gpu
+  build:
+    local:
+      push: true
   deploy:
     helm:
       releases: