|
| 1 | +name: rayjob-e2e-with-kueue |
| 2 | + |
| 3 | +on: |
| 4 | + pull_request: |
| 5 | + branches: |
| 6 | + - main |
| 7 | + - 'release-*' |
| 8 | + - ray-jobs-feature |
| 9 | + paths-ignore: |
| 10 | + - 'docs/**' |
| 11 | + - '**.adoc' |
| 12 | + - '**.md' |
| 13 | + - 'LICENSE' |
| 14 | + |
| 15 | +concurrency: |
| 16 | + group: ${{ github.head_ref }}-${{ github.workflow }} |
| 17 | + cancel-in-progress: true |
| 18 | + |
| 19 | +env: |
| 20 | + CODEFLARE_OPERATOR_IMG: "quay.io/project-codeflare/codeflare-operator:dev" |
| 21 | + KUEUE_VERSION: "v0.13.3" |
| 22 | + |
| 23 | +jobs: |
| 24 | + kubernetes: |
| 25 | + runs-on: ubuntu-latest |
| 26 | + |
| 27 | + steps: |
| 28 | + - name: Checkout code |
| 29 | + uses: actions/checkout@v4 |
| 30 | + with: |
| 31 | + submodules: recursive |
| 32 | + |
| 33 | + - name: Checkout common repo code |
| 34 | + uses: actions/checkout@v4 |
| 35 | + with: |
| 36 | + repository: 'project-codeflare/codeflare-common' |
| 37 | + ref: 'main' |
| 38 | + path: 'common' |
| 39 | + |
| 40 | + |
| 41 | + - name: Set up specific Python version |
| 42 | + uses: actions/setup-python@v5 |
| 43 | + with: |
| 44 | + python-version: '3.11' |
| 45 | + cache: 'pip' # caching pip dependencies |
| 46 | + |
| 47 | + - name: Setup and start KinD cluster |
| 48 | + uses: ./common/github-actions/kind |
| 49 | + with: |
| 50 | + worker-nodes: 2 # Multiple nodes for testing Kueue scheduling |
| 51 | + |
| 52 | + - name: Verify Kind cluster |
| 53 | + run: | |
| 54 | + echo "Checking Kind clusters..." |
| 55 | + kind get clusters |
| 56 | + echo "Current kubectl context:" |
| 57 | + kubectl config current-context |
| 58 | + echo "Checking nodes:" |
| 59 | + kubectl get nodes |
| 60 | +
|
| 61 | + - name: Deploy Kueue |
| 62 | + run: | |
| 63 | + echo "Deploying Kueue ${KUEUE_VERSION}" |
| 64 | + kubectl apply --server-side -f https://github.com/kubernetes-sigs/kueue/releases/download/${KUEUE_VERSION}/manifests.yaml |
| 65 | +
|
| 66 | + # Sleep until the kueue manager is running |
| 67 | + echo "Waiting for pods in the kueue-system namespace to become ready" |
| 68 | + while [[ $(kubectl get pods -n kueue-system -o 'jsonpath={..status.conditions[?(@.type=="Ready")].status}' | tr ' ' '\n' | sort -u) != "True" ]] |
| 69 | + do |
| 70 | + echo -n "." && sleep 1; |
| 71 | + done |
| 72 | + echo "" |
| 73 | +
|
| 74 | + sleep 5 |
| 75 | +
|
| 76 | + - name: Deploy KubeRay operator |
| 77 | + run: | |
| 78 | + KUBERAY_VERSION="v1.4.0" |
| 79 | + echo "Deploying KubeRay ${KUBERAY_VERSION}" |
| 80 | +
|
| 81 | + # Create namespace first |
| 82 | + kubectl create namespace ray-system || true |
| 83 | +
|
| 84 | + kubectl apply --server-side -k "github.com/ray-project/kuberay/ray-operator/config/default?ref=${KUBERAY_VERSION}&timeout=180s" |
| 85 | +
|
| 86 | + # Check all namespaces to see where KubeRay was deployed |
| 87 | + echo "Checking for KubeRay deployment in all namespaces..." |
| 88 | + kubectl get deployments -A | grep kuberay || true |
| 89 | +
|
| 90 | + # Wait for KubeRay operator to be ready |
| 91 | + echo "Waiting for KubeRay operator to become ready..." |
| 92 | + kubectl wait --for=condition=Available --timeout=300s deployment/kuberay-operator -n ray-system || { |
| 93 | + echo "KubeRay operator not found in ray-system, checking other namespaces:" |
| 94 | + kubectl get pods -A | grep kuberay |
| 95 | + kubectl get deployments -A | grep kuberay |
| 96 | + # Try default namespace |
| 97 | + kubectl wait --for=condition=Available --timeout=30s deployment/kuberay-operator -n default || true |
| 98 | + # Try kuberay-system namespace |
| 99 | + kubectl wait --for=condition=Available --timeout=30s deployment/kuberay-operator -n kuberay-system || true |
| 100 | + } |
| 101 | +
|
| 102 | + - name: Add user to KinD |
| 103 | + uses: ./common/github-actions/kind-add-user |
| 104 | + with: |
| 105 | + user-name: sdk-user |
| 106 | + |
| 107 | + - name: Configure RBAC for sdk user with limited permissions |
| 108 | + run: | |
| 109 | + # Basic permissions |
| 110 | + kubectl create clusterrole list-ingresses --verb=get,list --resource=ingresses |
| 111 | + kubectl create clusterrolebinding sdk-user-list-ingresses --clusterrole=list-ingresses --user=sdk-user |
| 112 | + kubectl create clusterrole namespace-creator --verb=get,list,create,delete,patch --resource=namespaces |
| 113 | + kubectl create clusterrolebinding sdk-user-namespace-creator --clusterrole=namespace-creator --user=sdk-user |
| 114 | +
|
| 115 | + # Ray permissions |
| 116 | + kubectl create clusterrole raycluster-creator --verb=get,list,create,delete,patch,watch --resource=rayclusters |
| 117 | + kubectl create clusterrolebinding sdk-user-raycluster-creator --clusterrole=raycluster-creator --user=sdk-user |
| 118 | + kubectl create clusterrole rayjob-creator --verb=get,list,create,delete,patch,watch,update --resource=rayjobs,rayjobs/status |
| 119 | + kubectl create clusterrolebinding sdk-user-rayjob-creator --clusterrole=rayjob-creator --user=sdk-user |
| 120 | +
|
| 121 | + # Kueue permissions |
| 122 | + kubectl create clusterrole resourceflavor-creator --verb=get,list,create,delete --resource=resourceflavors |
| 123 | + kubectl create clusterrolebinding sdk-user-resourceflavor-creator --clusterrole=resourceflavor-creator --user=sdk-user |
| 124 | + kubectl create clusterrole clusterqueue-creator --verb=get,list,create,delete,patch --resource=clusterqueues |
| 125 | + kubectl create clusterrolebinding sdk-user-clusterqueue-creator --clusterrole=clusterqueue-creator --user=sdk-user |
| 126 | + kubectl create clusterrole localqueue-creator --verb=get,list,create,delete,patch --resource=localqueues |
| 127 | + kubectl create clusterrolebinding sdk-user-localqueue-creator --clusterrole=localqueue-creator --user=sdk-user |
| 128 | + kubectl create clusterrole workload-creator --verb=get,list,watch --resource=workloads |
| 129 | + kubectl create clusterrolebinding sdk-user-workload-creator --clusterrole=workload-creator --user=sdk-user |
| 130 | +
|
| 131 | + # Additional permissions |
| 132 | + kubectl create clusterrole list-secrets --verb=get,list --resource=secrets |
| 133 | + kubectl create clusterrolebinding sdk-user-list-secrets --clusterrole=list-secrets --user=sdk-user |
| 134 | + kubectl create clusterrole pod-creator --verb=get,list,watch --resource=pods |
| 135 | + kubectl create clusterrolebinding sdk-user-pod-creator --clusterrole=pod-creator --user=sdk-user |
| 136 | + kubectl create clusterrole service-reader --verb=get,list,watch --resource=services |
| 137 | + kubectl create clusterrolebinding sdk-user-service-reader --clusterrole=service-reader --user=sdk-user |
| 138 | + kubectl create clusterrole port-forward-pods --verb=create --resource=pods/portforward |
| 139 | + kubectl create clusterrolebinding sdk-user-port-forward-pods-binding --clusterrole=port-forward-pods --user=sdk-user |
| 140 | + kubectl create clusterrole node-reader --verb=get,list --resource=nodes |
| 141 | + kubectl create clusterrolebinding sdk-user-node-reader --clusterrole=node-reader --user=sdk-user |
| 142 | + kubectl config use-context sdk-user |
| 143 | +
|
| 144 | + - name: Setup test output directory |
| 145 | + run: | |
| 146 | + CODEFLARE_TEST_OUTPUT_DIR="${{ runner.temp }}/test-logs" |
| 147 | + mkdir -p ${CODEFLARE_TEST_OUTPUT_DIR} |
| 148 | + echo "CODEFLARE_TEST_OUTPUT_DIR=${CODEFLARE_TEST_OUTPUT_DIR}" >> $GITHUB_ENV |
| 149 | +
|
| 150 | + - name: Run RayJob e2e tests |
| 151 | + run: | |
| 152 | + set -euo pipefail |
| 153 | + pip install poetry |
| 154 | + poetry install --with test,docs |
| 155 | +
|
| 156 | + # Install the SDK in editable mode |
| 157 | + pip install -e . |
| 158 | +
|
| 159 | + echo "Running RayJob e2e tests..." |
| 160 | + # Set environment variable to prevent default queue assignment for non-Kueue tests |
| 161 | + export DISABLE_DEFAULT_KUEUE_QUEUE=true |
| 162 | +
|
| 163 | + # Run only the tests that are designed for Kueue integration |
| 164 | + poetry run pytest -v -s ./tests/e2e/rayjob/rayjob_existing_cluster_test.py ./tests/e2e/rayjob/rayjob_lifecycled_cluster_test.py -x > ${CODEFLARE_TEST_OUTPUT_DIR}/pytest_output.log 2>&1 |
| 165 | + env: |
| 166 | + GRPC_DNS_RESOLVER: "native" |
| 167 | + |
| 168 | + - name: Switch to kind-cluster context to print logs |
| 169 | + if: always() |
| 170 | + run: kubectl config use-context kind-cluster |
| 171 | + |
| 172 | + - name: Print Pytest output log |
| 173 | + if: always() |
| 174 | + run: | |
| 175 | + echo "Printing Pytest output logs" |
| 176 | + cat ${{ env.CODEFLARE_TEST_OUTPUT_DIR }}/pytest_output.log || true |
| 177 | +
|
| 178 | + - name: Print Kueue operator logs |
| 179 | + if: always() |
| 180 | + run: | |
| 181 | + echo "Printing Kueue operator logs" |
| 182 | + kubectl logs -n kueue-system --tail -1 -l control-plane=controller-manager | tee ${{ env.CODEFLARE_TEST_OUTPUT_DIR }}/kueue-operator.log || true |
| 183 | +
|
| 184 | + - name: Print KubeRay operator logs |
| 185 | + if: always() |
| 186 | + run: | |
| 187 | + echo "Printing KubeRay operator logs" |
| 188 | + echo "Checking ray-system namespace contents:" |
| 189 | + kubectl get all -n ray-system || true |
| 190 | + echo "Attempting to get KubeRay logs with different selectors:" |
| 191 | + kubectl logs -n ray-system --tail -1 -l app.kubernetes.io/name=kuberay-operator | tee ${{ env.CODEFLARE_TEST_OUTPUT_DIR }}/kuberay.log || \ |
| 192 | + kubectl logs -n ray-system --tail -1 -l app.kubernetes.io/component=kuberay-operator | tee ${{ env.CODEFLARE_TEST_OUTPUT_DIR }}/kuberay.log || \ |
| 193 | + kubectl logs -n ray-system --tail -1 deployment/kuberay-operator | tee ${{ env.CODEFLARE_TEST_OUTPUT_DIR }}/kuberay.log || \ |
| 194 | + echo "Could not find KubeRay operator logs" |
| 195 | +
|
| 196 | + - name: Export all KinD pod logs |
| 197 | + uses: ./common/github-actions/kind-export-logs |
| 198 | + if: always() |
| 199 | + with: |
| 200 | + output-directory: ${{ env.CODEFLARE_TEST_OUTPUT_DIR }} |
| 201 | + |
| 202 | + - name: Upload logs |
| 203 | + uses: actions/upload-artifact@v4 |
| 204 | + if: always() |
| 205 | + with: |
| 206 | + name: logs |
| 207 | + retention-days: 10 |
| 208 | + path: | |
| 209 | + ${{ env.CODEFLARE_TEST_OUTPUT_DIR }}/**/*.log |
| 210 | + if-no-files-found: warn |
0 commit comments