[CI/Build] Add x86_64 and aarch64 docker image builds to release pipeline

Add release Docker image build steps to the nightly release pipeline
that build per-arch images from docker/Dockerfile.cuda, push them to
ECR, create a multi-arch manifest, and publish it to Docker Hub under
the nightly tag at vllm/vllm-omni.

Mirrors the pattern used by vllm-project/vllm's release pipeline.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

.buildkite/nightly-release-pipeline.yaml

@@ -18,3 +18,70 @@ steps:
       queue: small_cpu_queue_release
     commands:
       - "bash .buildkite/scripts/generate-and-upload-nightly-index.sh"
+
+  - group: "Build release Docker images"
+    key: "build-release-images"
+    steps:
+      - label: "Build release image - x86_64"
+        key: build-release-image-x86
+        depends_on: ~
+        agents:
+          queue: cpu_queue_release
+        commands:
+          - "aws ecr-public get-login-password --region us-east-1 | docker login --username AWS --password-stdin public.ecr.aws/q9t5s3a7"
+          - |
+            DOCKER_BUILDKIT=1 docker build \
+              --build-arg max_jobs=16 \
+              --tag public.ecr.aws/q9t5s3a7/vllm-omni-release-repo:$BUILDKITE_COMMIT-$(uname -m) \
+              --progress plain \
+              -f docker/Dockerfile.cuda .
+          - "docker push public.ecr.aws/q9t5s3a7/vllm-omni-release-repo:$BUILDKITE_COMMIT-$(uname -m)"
+          # re-tag to default image tag and push, just in case arm64 build fails
+          - "docker tag public.ecr.aws/q9t5s3a7/vllm-omni-release-repo:$BUILDKITE_COMMIT-$(uname -m) public.ecr.aws/q9t5s3a7/vllm-omni-release-repo:$BUILDKITE_COMMIT"
+          - "docker push public.ecr.aws/q9t5s3a7/vllm-omni-release-repo:$BUILDKITE_COMMIT"
+        env:
+          DOCKER_BUILDKIT: "1"
+
+      - label: "Build release image - aarch64"
+        key: build-release-image-arm64
+        depends_on: ~
+        agents:
+          queue: arm64_cpu_queue_release
+        commands:
+          - "aws ecr-public get-login-password --region us-east-1 | docker login --username AWS --password-stdin public.ecr.aws/q9t5s3a7"
+          - |
+            DOCKER_BUILDKIT=1 docker build \
+              --build-arg max_jobs=16 \
+              --tag public.ecr.aws/q9t5s3a7/vllm-omni-release-repo:$BUILDKITE_COMMIT-$(uname -m) \
+              --progress plain \
+              -f docker/Dockerfile.cuda .
+          - "docker push public.ecr.aws/q9t5s3a7/vllm-omni-release-repo:$BUILDKITE_COMMIT-$(uname -m)"
+        env:
+          DOCKER_BUILDKIT: "1"
+
+  - label: "Create multi-arch manifest"
+    key: create-multi-arch-manifest
+    depends_on:
+      - build-release-image-x86
+      - build-release-image-arm64
+    agents:
+      queue: small_cpu_queue_release
+    commands:
+      - "aws ecr-public get-login-password --region us-east-1 | docker login --username AWS --password-stdin public.ecr.aws/q9t5s3a7"
+      - "docker manifest create public.ecr.aws/q9t5s3a7/vllm-omni-release-repo:$BUILDKITE_COMMIT public.ecr.aws/q9t5s3a7/vllm-omni-release-repo:$BUILDKITE_COMMIT-x86_64 public.ecr.aws/q9t5s3a7/vllm-omni-release-repo:$BUILDKITE_COMMIT-aarch64 --amend"
+      - "docker manifest push public.ecr.aws/q9t5s3a7/vllm-omni-release-repo:$BUILDKITE_COMMIT"
+
+  - label: "Publish nightly multi-arch image to DockerHub"
+    depends_on:
+      - create-multi-arch-manifest
+    agents:
+      queue: small_cpu_queue_release
+    commands:
+      - "bash .buildkite/scripts/push-nightly-builds.sh"
+    plugins:
+      - docker-login#v3.0.0:
+          username: vllmbot
+          password-env: DOCKERHUB_TOKEN
+    env:
+      DOCKER_BUILDKIT: "1"
+      DOCKERHUB_USERNAME: "vllmbot"

.buildkite/scripts/push-nightly-builds.sh

@@ -0,0 +1,30 @@
+#!/bin/bash
+
+set -ex
+
+# Pull per-arch images from ECR (built earlier in the pipeline) and re-publish
+# them to Docker Hub under a nightly tag, then create and push a multi-arch
+# manifest pointing at both per-arch tags.
+
+ORIG_TAG_NAME="$BUILDKITE_COMMIT"
+TAG_NAME="nightly"
+DOCKERHUB_REPO="vllm/vllm-omni"
+ECR_REPO="public.ecr.aws/q9t5s3a7/vllm-omni-release-repo"
+
+echo "Pushing original tag $ORIG_TAG_NAME to new nightly tag name: $TAG_NAME"
+
+# pull original arch-dependent images from AWS ECR Public
+aws ecr-public get-login-password --region us-east-1 | docker login --username AWS --password-stdin public.ecr.aws/q9t5s3a7
+docker pull "$ECR_REPO:$ORIG_TAG_NAME-x86_64"
+docker pull "$ECR_REPO:$ORIG_TAG_NAME-aarch64"
+# tag arch-dependent images
+docker tag "$ECR_REPO:$ORIG_TAG_NAME-x86_64" "$DOCKERHUB_REPO:$TAG_NAME-x86_64"
+docker tag "$ECR_REPO:$ORIG_TAG_NAME-aarch64" "$DOCKERHUB_REPO:$TAG_NAME-aarch64"
+# push arch-dependent images to DockerHub
+docker push "$DOCKERHUB_REPO:$TAG_NAME-x86_64"
+docker push "$DOCKERHUB_REPO:$TAG_NAME-aarch64"
+# push arch-independent manifest to DockerHub
+docker manifest create "$DOCKERHUB_REPO:$TAG_NAME" "$DOCKERHUB_REPO:$TAG_NAME-x86_64" "$DOCKERHUB_REPO:$TAG_NAME-aarch64" --amend
+docker manifest create "$DOCKERHUB_REPO:$TAG_NAME-$BUILDKITE_COMMIT" "$DOCKERHUB_REPO:$TAG_NAME-x86_64" "$DOCKERHUB_REPO:$TAG_NAME-aarch64" --amend
+docker manifest push "$DOCKERHUB_REPO:$TAG_NAME"
+docker manifest push "$DOCKERHUB_REPO:$TAG_NAME-$BUILDKITE_COMMIT"