Ci new cluster

.github/workflows/build.yaml

@@ -0,0 +1,164 @@
+name: Build and push docker image to registry
+
+on:
+  workflow_dispatch:
+  push:
+    branches:
+      - 'main'
+    tags:
+      - 'v*'
+  pull_request:
+    paths:
+      - ".github/workflows/build.yaml"
+      - ".github/workflows/matrix.json"
+      - "integration-tests/**"
+      - "backends/**"
+      - "core/**"
+      - "router/**"
+      - "Cargo.lock"
+      - "rust-toolchain.toml"
+      - "Dockerfile"
+    branches:
+      - 'main'
+
+jobs:
+  matrix:
+    runs-on: ubuntu-latest
+    outputs:
+      matrix: ${{ steps.set-matrix.outputs.matrix }}
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@v3
+
+      - id: set-matrix
+        run: |
+          branchName=$(echo '${{ github.ref }}' | sed 's,refs/heads/,,g')
+          matrix=$(jq --arg branchName "$branchName" 'map(. | select((.runOn==$branchName) or (.runOn=="always")) )' .github/workflows/matrix.json)
+          echo "{\"include\":$(echo $matrix)}"
+          echo ::set-output name=matrix::{\"include\":$(echo $matrix)}\"
+
+  build-and-push-image:
+    needs: matrix
+    strategy:
+      matrix: ${{fromJson(needs.matrix.outputs.matrix)}}
+    concurrency:
+      group: ${{ github.workflow }}-${{ github.job }}-${{matrix.name}}-${{ github.head_ref || github.run_id }}
+      cancel-in-progress: true
+    runs-on:
+      group: aws-r7i-8xlarge-priv
+    permissions:
+      contents: write
+      packages: write
+      # This is used to complete the identity challenge
+      # with sigstore/fulcio when running outside of PRs.
+      id-token: write
+      security-events: write
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@v4
+
+      - name: Initialize Docker Buildx
+        uses: docker/setup-buildx-action@v3
+        with:
+          install: true
+          buildkitd-config-inline: |
+            [registry."docker.io"]
+              mirrors = ["registry-us-east-1.prod.aws.ci.huggingface.tech"]
+
+      - name: Configure sccache
+        uses: actions/github-script@v6
+        with:
+          script: |
+            core.exportVariable('ACTIONS_CACHE_URL', process.env.ACTIONS_CACHE_URL || '');
+            core.exportVariable('ACTIONS_RUNTIME_TOKEN', process.env.ACTIONS_RUNTIME_TOKEN || '');
+
+      - name: Inject slug/short variables
+        uses: rlespinasse/github-slug-action@v4
+
+      - name: Login to internal Container Registry
+        if: github.event_name != 'pull_request'
+        uses: docker/login-action@v3
+        with:
+          username: ${{ secrets.REGISTRY_USERNAME }}
+          password: ${{ secrets.REGISTRY_PASSWORD }}
+          registry: registry.internal.huggingface.tech
+
+      - name: Login to GitHub Container Registry
+        if: github.event_name != 'pull_request'
+        uses: docker/login-action@v3
+        with:
+          registry: ghcr.io
+          username: ${{ github.actor }}
+          password: ${{ secrets.GITHUB_TOKEN }}
+
+      - name: Extract metadata (tags, labels) for Docker
+        id: meta
+        uses: docker/metadata-action@v5
+        with:
+          images: |
+            registry.internal.huggingface.tech/api-inference/text-embeddings-inference
+            ghcr.io/huggingface/text-embeddings-inference
+          flavor: |
+            latest=false
+          tags: |
+            type=semver,pattern=${{ matrix.imageNamePrefix }}{{version}}
+            type=semver,pattern=${{ matrix.imageNamePrefix }}{{major}}.{{minor}}
+            type=raw,value=${{ matrix.imageNamePrefix }}latest
+            type=raw,value=${{ matrix.imageNamePrefix }}sha-${{ env.GITHUB_SHA_SHORT }}
+
+      - name: Build and push Docker image
+        id: build-and-push
+        uses: docker/build-push-action@v6
+        with:
+          context: .
+          file: ${{ matrix.dockerfile }}
+          push: ${{ github.event_name != 'pull_request' }}
+          platforms: 'linux/amd64'
+          build-args: |
+            SCCACHE_GHA_ENABLED=${{ matrix.sccache }}
+            ACTIONS_CACHE_URL=${{ env.ACTIONS_CACHE_URL }}
+            ACTIONS_RUNTIME_TOKEN=${{ env.ACTIONS_RUNTIME_TOKEN }}
+            CUDA_COMPUTE_CAP=${{ matrix.cudaComputeCap }}
+            GIT_SHA=${{ env.GITHUB_SHA }}
+            DOCKER_LABEL=sha-${{ env.GITHUB_SHA_SHORT }}
+            ${{matrix.extraBuildArgs}}
+          tags: ${{ steps.meta.outputs.tags }}
+          labels: ${{ steps.meta.outputs.labels }}
+          cache-from: type=s3,region=us-east-1,bucket=ci-docker-buildx-cache,name=text-embeddings-inference-cache-${{matrix.name}},access_key_id=${{ secrets.S3_CI_DOCKER_BUILDX_CACHE_ACCESS_KEY_ID }},secret_access_key=${{ secrets.S3_CI_DOCKER_BUILDX_CACHE_SECRET_ACCESS_KEY }},mode=max
+          cache-to: type=s3,region=us-east-1,bucket=ci-docker-buildx-cache,name=text-embeddings-inference-cache-${{matrix.name}},access_key_id=${{ secrets.S3_CI_DOCKER_BUILDX_CACHE_ACCESS_KEY_ID }},secret_access_key=${{ secrets.S3_CI_DOCKER_BUILDX_CACHE_SECRET_ACCESS_KEY }},mode=max
+
+      - name: Extract metadata (tags, labels) for Docker
+        id: meta-grpc
+        uses: docker/metadata-action@v5
+        with:
+          images: |
+            registry.internal.huggingface.tech/api-inference/text-embeddings-inference
+            ghcr.io/huggingface/text-embeddings-inference
+          flavor: |
+            latest=false
+          tags: |
+            type=semver,pattern=${{ matrix.imageNamePrefix }}{{version}}-grpc
+            type=semver,pattern=${{ matrix.imageNamePrefix }}{{major}}.{{minor}}-grpc
+            type=raw,value=${{ matrix.imageNamePrefix }}latest-grpc
+            type=raw,value=${{ matrix.imageNamePrefix }}sha-${{ env.GITHUB_SHA_SHORT }}-grpc
+
+      - name: Build and push Docker image
+        id: build-and-push-grpc
+        uses: docker/build-push-action@v6
+        with:
+          context: .
+          target: grpc
+          file: Dockerfile-cuda
+          push: ${{ github.event_name != 'pull_request' }}
+          platforms: 'linux/amd64'
+          build-args: |
+            SCCACHE_GHA_ENABLED=${{ matrix.sccache }}
+            ACTIONS_CACHE_URL=${{ env.ACTIONS_CACHE_URL }}
+            ACTIONS_RUNTIME_TOKEN=${{ env.ACTIONS_RUNTIME_TOKEN }}
+            CUDA_COMPUTE_CAP=${{ matrix.cudaComputeCap }}
+            GIT_SHA=${{ env.GITHUB_SHA }}
+            DOCKER_LABEL=sha-${{ env.GITHUB_SHA_SHORT }}
+            ${{matrix.extraBuildArgs}}
+          tags: ${{ steps.meta-grpc.outputs.tags }}
+          labels: ${{ steps.meta-grpc.outputs.labels }}
+          cache-from: type=s3,region=us-east-1,bucket=ci-docker-buildx-cache,name=text-embeddings-inference-cache-${{matrix.name}},access_key_id=${{ secrets.S3_CI_DOCKER_BUILDX_CACHE_ACCESS_KEY_ID }},secret_access_key=${{ secrets.S3_CI_DOCKER_BUILDX_CACHE_SECRET_ACCESS_KEY }},mode=max