[CI] refine more logic when generating and using nightly wheels & indices, add cuda130 build for aarch64, specify correct manylinux version (#30341)

Signed-off-by: Shengqi Chen <harry-chen@outlook.com>

.buildkite/release-pipeline.yaml

@@ -15,6 +15,21 @@ steps:
     env:
       DOCKER_BUILDKIT: "1"
 
+  - label: "Build arm64 wheel - CUDA 13.0"
+    depends_on: ~
+    id: build-wheel-arm64-cuda-13-0
+    agents:
+      queue: arm64_cpu_queue_postmerge
+    commands:
+      # #NOTE: torch_cuda_arch_list is derived from upstream PyTorch build files here:
+      # https://github.com/pytorch/pytorch/blob/main/.ci/aarch64_linux/aarch64_ci_build.sh#L7
+      - "DOCKER_BUILDKIT=1 docker build --build-arg max_jobs=16 --build-arg USE_SCCACHE=1 --build-arg GIT_REPO_CHECK=1 --build-arg CUDA_VERSION=13.0.1 --build-arg torch_cuda_arch_list='8.7 8.9 9.0 10.0+PTX 12.0' --build-arg BUILD_BASE_IMAGE=nvidia/cuda:13.0.1-devel-ubuntu22.04  --tag vllm-ci:build-image --target build --progress plain -f docker/Dockerfile ."
+      - "mkdir artifacts"
+      - "docker run --rm -v $(pwd)/artifacts:/artifacts_host vllm-ci:build-image bash -c 'cp -r dist /artifacts_host && chmod -R a+rw /artifacts_host'"
+      - "bash .buildkite/scripts/upload-wheels.sh manylinux_2_35"
+    env:
+      DOCKER_BUILDKIT: "1"
+
   # aarch64 build
   - label: "Build arm64 CPU wheel"
     depends_on: ~
@@ -25,7 +40,7 @@ steps:
       - "DOCKER_BUILDKIT=1 docker build --build-arg max_jobs=16 --build-arg GIT_REPO_CHECK=1 --build-arg VLLM_BUILD_ACL=ON --tag vllm-ci:build-image --target vllm-build --progress plain -f docker/Dockerfile.cpu ."
       - "mkdir artifacts"
       - "docker run --rm -v $(pwd)/artifacts:/artifacts_host vllm-ci:build-image bash -c 'cp -r dist /artifacts_host && chmod -R a+rw /artifacts_host'"
-      - "bash .buildkite/scripts/upload-wheels.sh"
+      - "bash .buildkite/scripts/upload-wheels.sh manylinux_2_35"
     env:
       DOCKER_BUILDKIT: "1"
 
@@ -39,7 +54,7 @@ steps:
       - "DOCKER_BUILDKIT=1 docker build --build-arg max_jobs=16 --build-arg USE_SCCACHE=1 --build-arg GIT_REPO_CHECK=1 --build-arg CUDA_VERSION=12.9.1 --tag vllm-ci:build-image --target build --progress plain -f docker/Dockerfile ."
       - "mkdir artifacts"
       - "docker run --rm -v $(pwd)/artifacts:/artifacts_host vllm-ci:build-image bash -c 'cp -r dist /artifacts_host && chmod -R a+rw /artifacts_host'"
-      - "bash .buildkite/scripts/upload-wheels.sh"
+      - "bash .buildkite/scripts/upload-wheels.sh manylinux_2_31"
     env:
       DOCKER_BUILDKIT: "1"
 
@@ -52,7 +67,7 @@ steps:
       - "DOCKER_BUILDKIT=1 docker build --build-arg max_jobs=16 --build-arg USE_SCCACHE=1 --build-arg GIT_REPO_CHECK=1 --build-arg CUDA_VERSION=13.0.1 --build-arg BUILD_BASE_IMAGE=nvidia/cuda:13.0.1-devel-ubuntu22.04 --tag vllm-ci:build-image --target build --progress plain -f docker/Dockerfile ."
       - "mkdir artifacts"
       - "docker run --rm -v $(pwd)/artifacts:/artifacts_host vllm-ci:build-image bash -c 'cp -r dist /artifacts_host && chmod -R a+rw /artifacts_host'"
-      - "bash .buildkite/scripts/upload-wheels.sh"
+      - "bash .buildkite/scripts/upload-wheels.sh manylinux_2_35"
     env:
       DOCKER_BUILDKIT: "1"
 

.buildkite/scripts/generate-nightly-index.py

@@ -372,6 +372,17 @@ if __name__ == "__main__":
 
     print(f"Found {len(wheel_files)} wheel files for version {version}: {wheel_files}")
 
+    # keep only "official" files for a non-nightly version (specifed by cli args)
+    PY_VERSION_RE = re.compile(r"^\d+\.\d+\.\d+([a-zA-Z0-9.+-]*)?$")
+    if PY_VERSION_RE.match(version):
+        # upload-wheels.sh ensures no "dev" is in args.version
+        wheel_files = list(
+            filter(lambda x: version in x and "dev" not in x, wheel_files)
+        )
+        print(f"Non-nightly version detected, wheel files used: {wheel_files}")
+    else:
+        print("Nightly version detected, keeping all wheel files.")
+
     # Generate index and metadata, assuming wheels and indices are stored as:
     # s3://vllm-wheels/{version}/<wheel files>
     # s3://vllm-wheels/<anything>/<index files>

.buildkite/scripts/upload-wheels.sh

@@ -34,9 +34,10 @@ if [[ ${#wheel_files[@]} -ne 1 ]]; then
 fi
 wheel="${wheel_files[0]}"
 
-# current build image uses ubuntu 20.04, which corresponds to manylinux_2_31
+# default build image uses ubuntu 20.04, which corresponds to manylinux_2_31
+# we also accept params as manylinux tag
 # refer to https://github.com/mayeut/pep600_compliance?tab=readme-ov-file#acceptable-distros-to-build-wheels
-manylinux_version="manylinux_2_31"
+manylinux_version="${1:-manylinux_2_31}"
 
 # Rename 'linux' to the appropriate manylinux version in the wheel filename
 if [[ "$wheel" != *"linux"* ]]; then
@@ -96,8 +97,11 @@ if [[ "$BUILDKITE_BRANCH" == "main" && "$BUILDKITE_PULL_REQUEST" == "false" ]];
     aws s3 cp --recursive "$INDICES_OUTPUT_DIR/" "s3://$BUCKET/nightly/"
 fi
 
-# copy to /<pure_version>/ only if it does not have "dev" in the version
+# re-generate and copy to /<pure_version>/ only if it does not have "dev" in the version
 if [[ "$version" != *"dev"* ]]; then
-    echo "Uploading indices to overwrite /$pure_version/"
+    echo "Re-generating indices for /$pure_version/"
+    rm -rf "$INDICES_OUTPUT_DIR/*"
+    mkdir -p "$INDICES_OUTPUT_DIR"
+    $PYTHON .buildkite/scripts/generate-nightly-index.py --version "$pure_version" --current-objects "$obj_json" --output-dir "$INDICES_OUTPUT_DIR" --comment "version $pure_version" $alias_arg
     aws s3 cp --recursive "$INDICES_OUTPUT_DIR/" "s3://$BUCKET/$pure_version/"
 fi

tests/standalone_tests/python_only_compile.sh

@@ -3,12 +3,45 @@
 # for users who do not have any compilers installed on their system
 
 set -e
-set -x
 
 merge_base_commit=$(git merge-base HEAD origin/main)
-echo "Current merge base commit with main: $merge_base_commit"
+echo "INFO: current merge base commit with main: $merge_base_commit"
 git show --oneline -s $merge_base_commit
 
+# test whether the metadata.json url is valid, retry each 3 minutes up to 5 times
+# this avoids cumbersome error messages & manual retries in case the precompiled wheel
+# for the given commit is still being built in the release pipeline
+meta_json_url="https://wheels.vllm.ai/$merge_base_commit/vllm/metadata.json"
+echo "INFO: will use metadata.json from $meta_json_url"
+
+for i in {1..5}; do
+    echo "Checking metadata.json URL (attempt $i)..."
+    if curl --fail "$meta_json_url" > metadata.json; then
+        echo "INFO: metadata.json URL is valid."
+        # check whether it is valid json by python
+        if python3 -m json.tool metadata.json; then
+            echo "INFO: metadata.json is valid JSON. Proceeding with the test."
+        else
+            echo "CRITICAL: metadata.json exists but is not valid JSON, please do report in #sig-ci channel!"
+            exit 1
+        fi
+        break
+    fi
+    # failure handling
+    if [ $i -eq 5 ]; then
+        echo "ERROR: metadata.json URL is still not valid after 5 attempts."
+        echo "ERROR: Please check whether the precompiled wheel for commit $merge_base_commit exists."
+        echo " NOTE: If $merge_base_commit is a new commit on main, maybe try again after its release pipeline finishes."
+        echo " NOTE: If it fails, please report in #sig-ci channel."
+        exit 1
+    else
+        echo "WARNING: metadata.json URL is not valid. Retrying in 3 minutes..."
+        sleep 180
+    fi
+done
+
+set -x
+
 cd /vllm-workspace/
 
 # uninstall vllm
@@ -29,6 +62,6 @@ python3 -c 'import vllm'
 
 # Check if the clangd log file was created
 if [ ! -f /tmp/changed.file ]; then
-    echo "changed.file was not created, python only compilation failed"
+    echo "ERROR: changed.file was not created, python only compilation failed"
     exit 1
 fi