fix: Add resource requirements #560
Workflow file for this run
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| name: e2e-test | |
| concurrency: | |
| group: ${{ github.workflow }}-${{ github.head_ref || github.run_id }} | |
| cancel-in-progress: true | |
| on: | |
| push: | |
| branches: [main] | |
| paths-ignore: ['docs/**', '**.md', '**.mdx', '**.png', '**.jpg'] | |
| pull_request: | |
| branches: [main] | |
| paths-ignore: ['docs/**', '**.md', '**.mdx', '**.png', '**.jpg'] | |
| env: | |
| GO_VERSION: "1.20" | |
| E2E_IMG_TAG: "kaito-e2e" | |
| permissions: | |
| id-token: write # This is required for requesting the JWT | |
| contents: read # This is required for actions/checkout | |
| jobs: | |
| e2e-tests: | |
| runs-on: ubuntu-latest | |
| environment: e2e-test | |
| steps: | |
| - name: Set up Go ${{ env.GO_VERSION }} | |
| uses: actions/setup-go@v5 | |
| with: | |
| go-version: ${{ env.GO_VERSION }} | |
| - name: Checkout | |
| uses: actions/checkout@v4 | |
| with: | |
| submodules: true | |
| fetch-depth: 0 | |
| - name: Shorten SHA | |
| id: vars | |
| run: echo "pr_sha_short=$(git rev-parse --short ${{ github.event.pull_request.head.sha }})" >> $GITHUB_OUTPUT | |
| - name: Set e2e Resource and Cluster Name | |
| run: | | |
| rand=${{ steps.vars.outputs.pr_sha_short }} | |
| if [ "$rand" = "" ]; then | |
| rand=$RANDOM | |
| fi | |
| echo "VERSION=${rand}" >> $GITHUB_ENV | |
| echo "CLUSTER_NAME=kaito${rand}" >> $GITHUB_ENV | |
| echo "RUN_LLAMA_13B=false" >> $GITHUB_ENV | |
| echo "AI_MODELS_IMAGE_VERSION=0.0.1" >> $GITHUB_ENV | |
| - uses: azure/[email protected] | |
| with: | |
| client-id: ${{ secrets.AZURE_CLIENT_ID }} | |
| tenant-id: ${{ secrets.AZURE_TENANT_ID }} | |
| subscription-id: ${{ secrets.AZURE_SUBSCRIPTION_ID }} | |
| - uses: azure/setup-helm@v3 | |
| with: | |
| token: ${{ secrets.GITHUB_TOKEN }} | |
| id: install | |
| - name: Create ACR | |
| shell: bash | |
| run: | | |
| make create-rg | |
| make create-acr | |
| env: | |
| AZURE_RESOURCE_GROUP: ${{ env.CLUSTER_NAME }} | |
| AZURE_ACR_NAME: ${{ env.CLUSTER_NAME }} | |
| - name: Create Azure Identity | |
| uses: azure/[email protected] | |
| with: | |
| inlineScript: | | |
| az identity create --name gpuIdentity --resource-group ${{ env.CLUSTER_NAME }} | |
| - name: build KAITO image | |
| shell: bash | |
| run: | | |
| make docker-build-kaito | |
| env: | |
| REGISTRY: ${{ env.CLUSTER_NAME }}.azurecr.io | |
| VERSION: ${{ env.VERSION }} | |
| - name: create cluster | |
| shell: bash | |
| run: | | |
| make create-aks-cluster | |
| env: | |
| AZURE_ACR_NAME: ${{ env.CLUSTER_NAME }} | |
| AZURE_RESOURCE_GROUP: ${{ env.CLUSTER_NAME }} | |
| AZURE_CLUSTER_NAME: ${{ env.CLUSTER_NAME }} | |
| - name: Install gpu-provisioner helm chart | |
| shell: bash | |
| run: | | |
| make gpu-provisioner-helm | |
| kubectl wait --for=condition=available deploy "kaito-gpu-provisioner" -n gpu-provisioner --timeout=300s | |
| env: | |
| AZURE_RESOURCE_GROUP: ${{ env.CLUSTER_NAME }} | |
| AZURE_CLUSTER_NAME: ${{ env.CLUSTER_NAME }} | |
| - uses: azure/[email protected] | |
| with: | |
| client-id: ${{ secrets.AZURE_CLIENT_ID }} | |
| tenant-id: ${{ secrets.AZURE_TENANT_ID }} | |
| subscription-id: ${{ secrets.AZURE_SUBSCRIPTION_ID }} | |
| - name: Create Role Assignment | |
| uses: azure/[email protected] | |
| with: | |
| inlineScript: | | |
| IDENTITY_PRINCIPAL_ID="$(az identity show --name gpuIdentity --resource-group ${{ env.CLUSTER_NAME }} --query 'principalId' -otsv)" | |
| az role assignment create --assignee ${IDENTITY_PRINCIPAL_ID} --scope "/subscriptions/${{ secrets.AZURE_SUBSCRIPTION_ID }}/resourceGroups/${{ env.CLUSTER_NAME }}" --role "Contributor" | |
| - name: Create Azure Federated Identity | |
| uses: azure/[email protected] | |
| with: | |
| inlineScript: | | |
| AKS_OIDC_ISSUER="$(az aks show -n "${{ env.CLUSTER_NAME }}" -g "${{ env.CLUSTER_NAME }}" --query 'oidcIssuerProfile.issuerUrl' -otsv)" | |
| az identity federated-credential create --name gpu-fed-credential --identity-name gpuIdentity --resource-group "${{ env.CLUSTER_NAME }}" \ | |
| --issuer "${AKS_OIDC_ISSUER}" --subject system:serviceaccount:"gpu-provisioner:gpu-provisioner" --audience api://AzureADTokenExchange | |
| - name: Install KAITO Workspace helm chart | |
| shell: bash | |
| run: | | |
| make az-patch-install-helm | |
| kubectl wait --for=condition=available deploy "kaito-workspace" -n kaito-workspace --timeout=300s | |
| env: | |
| AZURE_RESOURCE_GROUP: ${{ env.CLUSTER_NAME }} | |
| AZURE_CLUSTER_NAME: ${{ env.CLUSTER_NAME }} | |
| REGISTRY: ${{ env.CLUSTER_NAME }}.azurecr.io | |
| VERSION: ${{ env.VERSION }} | |
| - name: Login to Private Presets ACR | |
| uses: azure/[email protected] | |
| with: | |
| inlineScript: | | |
| az acr login --name ${{ secrets.DOCKER_SERVER }} --expose-token | |
| az aks update -n ${{ env.CLUSTER_NAME }} -g ${{ env.CLUSTER_NAME }} --attach-acr ${{secrets.ACR_AIMODELSREGISTRY}} | |
| - name: Add Secret Credentials | |
| run: | | |
| kubectl create secret docker-registry ${{secrets.DOCKER_REGISTRY}} \ | |
| --docker-server=${{secrets.DOCKER_SERVER}} \ | |
| --docker-username=${{secrets.DOCKER_USERNAME}} \ | |
| --docker-password=${{secrets.DOCKER_PASSWORD}} | |
| - name: Log kaito-workspace | |
| run: | | |
| kubectl get pods -n kaito-workspace -o name | grep "^pod/kaito-workspace" | sed 's/^pod\///' | xargs -I {} kubectl logs -n kaito-workspace {} | |
| - name: Run e2e test | |
| run: | | |
| make kaito-workspace-e2e-test | |
| env: | |
| AZURE_CLUSTER_NAME: ${{ env.CLUSTER_NAME }} | |
| RUN_LLAMA_13B: ${{ env.RUN_LLAMA_13B }} | |
| AI_MODELS_REGISTRY: ${{secrets.DOCKER_SERVER}} | |
| AI_MODELS_REGISTRY_SECRET: ${{ secrets.DOCKER_REGISTRY }} | |
| AI_MODELS_IMAGE_VERSION: ${{ env.AI_MODELS_IMAGE_VERSION }} | |
| - name: Cleanup e2e resources | |
| if: ${{ always() }} | |
| uses: azure/[email protected] | |
| with: | |
| inlineScript: | | |
| set +e | |
| az group delete --name "${{ env.CLUSTER_NAME }}" --yes --no-wait || true |