feat: Add manylinux support for Dynamo (#536)

Signed-off-by: Pavithra Vijayakrishnan <160681768+pvijayakrish@users.noreply.github.com>

container/Dockerfile.tensorrt_llm

@@ -15,8 +15,14 @@
 
 ARG BASE_IMAGE="tensorrt_llm/release"
 ARG BASE_IMAGE_TAG="latest"
+ARG MANYLINUX_IMAGE="quay.io/pypa/manylinux_2_28_x86_64"
+ARG RELEASE_BUILD
 
-FROM ${BASE_IMAGE}:${BASE_IMAGE_TAG} AS dev
+##################################
+########## Build Image ###########
+##################################
+
+FROM ${BASE_IMAGE}:${BASE_IMAGE_TAG} AS build
 
 USER root
 
@@ -116,15 +122,61 @@ COPY deploy/dynamo/sdk /workspace/deploy/dynamo/sdk
 
 # Install uv, create virtualenv for general use, and build dynamo wheel
 COPY --from=ghcr.io/astral-sh/uv:latest /uv /uvx /bin/
+
+### VIRTUAL ENVIRONMENT SETUP ###
+RUN mkdir /opt/dynamo && \
+    uv venv /opt/dynamo/venv --python 3.12
+
+###################################
+####### WHEEL BUILD STAGE #########
+###################################
+
+# Build the wheel in the manylinux environment
+FROM ${MANYLINUX_IMAGE} AS wheel_builder
+ARG RELEASE_BUILD
+ARG CARGO_BUILD_JOBS
+# Set CARGO_BUILD_JOBS to 16 if not provided
+# This is to prevent cargo from building $(nproc) jobs in parallel,
+# which might exceed the number of opened files limit.
+ENV CARGO_BUILD_JOBS=${CARGO_BUILD_JOBS:-16}
+
+WORKDIR /workspace
+
+RUN yum install -y protobuf-compiler \
+    && yum clean all \
+    && rm -rf /var/cache/yum
+
+COPY --from=build /workspace /workspace
+
+ENV RUSTUP_HOME=/usr/local/rustup \
+    CARGO_HOME=/usr/local/cargo \
+    PATH=/usr/local/cargo/bin:$PATH \
+    CARGO_TARGET_DIR=/workspace/target
+
+COPY --from=build $RUSTUP_HOME $RUSTUP_HOME
+COPY --from=build $CARGO_HOME $CARGO_HOME
+
+# Create virtualenv and build dynamo wheel
 RUN mkdir /opt/dynamo && \
     uv venv /opt/dynamo/venv --python 3.12 && \
     source /opt/dynamo/venv/bin/activate && \
     cd /workspace/lib/bindings/python && \
-    uv build --wheel --out-dir /workspace/dist && \
-    uv pip install /workspace/dist/ai_dynamo_runtime*cp312*.whl && \
+    uv build --wheel --out-dir /workspace/dist --python 3.12 && \
+    if [ "$RELEASE_BUILD" = "true" ]; then \
+        uv build --wheel --out-dir /workspace/dist --python 3.11 && \
+        uv build --wheel --out-dir /workspace/dist --python 3.10; \
+    fi && \
     cd /workspace && \
-    uv build --wheel --out-dir /workspace/dist && \
-    uv pip install /workspace/dist/ai_dynamo*any.whl
+    uv build --wheel --out-dir /workspace/dist
+
+########################################
+########## Development Image ###########
+########################################
+
+FROM build AS dev
+
+WORKDIR /workspace
+COPY --from=wheel_builder /workspace/dist/ /workspace/dist/
 
 # Package the bindings
 RUN mkdir -p /opt/dynamo/bindings/wheels && \
@@ -133,6 +185,10 @@ RUN mkdir -p /opt/dynamo/bindings/wheels && \
     cp target/release/libdynamo_llm_capi.so /opt/dynamo/bindings/lib/. && \
     cp -r lib/bindings/c/include /opt/dynamo/bindings/.
 
+RUN . /opt/dynamo/venv/bin/activate && \
+    uv pip install /workspace/dist/ai_dynamo_runtime*cp312*.whl && \
+    uv pip install /workspace/dist/ai_dynamo*any.whl
+
 # Install dynamo.runtime and dynamo.llm wheels globally in container for tests that
 # currently run without virtual environment activated.
 # TODO: In future, we may use a virtualenv for everything and remove this.

container/Dockerfile.vllm

@@ -3,10 +3,10 @@
 
 ARG BASE_IMAGE="nvcr.io/nvidia/cuda-dl-base"
 ARG BASE_IMAGE_TAG="25.01-cuda12.8-devel-ubuntu24.04"
-
+ARG RELEASE_BUILD
 ARG RUNTIME_IMAGE="nvcr.io/nvidia/cuda"
 ARG RUNTIME_IMAGE_TAG="12.8.1-runtime-ubuntu24.04"
-
+ARG MANYLINUX_IMAGE="quay.io/pypa/manylinux_2_28_x86_64"
 ARG GENAI_PERF_TAG="25d0188713adc47868d6b3f22426375237a90529"
 
 FROM ${BASE_IMAGE}:${BASE_IMAGE_TAG} AS nixl_base
@@ -280,24 +280,6 @@ RUN cargo build --release --locked --features mistralrs,sglang,vllm,python && \
 
 COPY deploy/dynamo/sdk /workspace/deploy/dynamo/sdk
 COPY deploy/dynamo/api-store /workspace/deploy/dynamo/api-store
-# Build dynamo wheel
-RUN source /opt/dynamo/venv/bin/activate && \
-    cd /workspace/lib/bindings/python && \
-    uv build --wheel --out-dir /workspace/dist && \
-    uv pip install /workspace/dist/ai_dynamo_runtime*cp312*.whl && \
-    cd /workspace && \
-    uv build --wheel --out-dir /workspace/dist && \
-    uv pip install /workspace/dist/ai_dynamo*any.whl && \
-    cd /workspace/deploy/dynamo/api-store && \
-    uv build --wheel --out-dir /workspace/dist && \
-    uv pip install /workspace/dist/ai_dynamo_store*any.whl
-
-# Package the bindings
-RUN mkdir -p /opt/dynamo/bindings/wheels && \
-    mkdir /opt/dynamo/bindings/lib && \
-    cp dist/ai_dynamo*cp312*.whl /opt/dynamo/bindings/wheels/. && \
-    cp target/release/libdynamo_llm_capi.so /opt/dynamo/bindings/lib/. && \
-    cp -r lib/bindings/c/include /opt/dynamo/bindings/.
 
 # Tell vllm to use the Dynamo LLM C API for KV Cache Routing
 ENV VLLM_KV_CAPI_PATH="/opt/dynamo/bindings/lib/libdynamo_llm_capi.so"
@@ -309,12 +291,74 @@ RUN --mount=type=bind,source=./container/launch_message.txt,target=/workspace/la
 
 CMD []
 
+###################################
+####### WHEEL BUILD STAGE #########
+###################################
+
+# Build the wheel in the manylinux environment
+FROM ${MANYLINUX_IMAGE} AS wheel_builder
+ARG CARGO_BUILD_JOBS
+# Set CARGO_BUILD_JOBS to 16 if not provided
+# This is to prevent cargo from building $(nproc) jobs in parallel,
+# which might exceed the number of opened files limit.
+ENV CARGO_BUILD_JOBS=${CARGO_BUILD_JOBS:-16}
+# Use build arg RELEASE_BUILD = true to generate wheels for Python 3.10, 3.11 and 3.12.
+ARG RELEASE_BUILD
+WORKDIR /workspace
+
+RUN yum install -y  protobuf-compiler \
+  && yum clean all \
+  && rm -rf /var/cache/yum
+
+ENV RUSTUP_HOME=/usr/local/rustup \
+    CARGO_HOME=/usr/local/cargo \
+    PATH=/usr/local/cargo/bin:$PATH \
+    CARGO_TARGET_DIR=/workspace/target
+
+COPY --from=build /workspace /workspace
+COPY --from=build $RUSTUP_HOME $RUSTUP_HOME
+COPY --from=build $CARGO_HOME $CARGO_HOME
+
+# Copy uv from build and build wheel in virtualenv
+RUN mkdir /opt/dynamo && \
+    uv venv /opt/dynamo/venv --python 3.12
+
+# Activate virtual environment
+ENV VIRTUAL_ENV=/opt/dynamo/venv
+ENV PATH="${VIRTUAL_ENV}/bin:${PATH}"
+ENV PATH="$PATH:/usr/local/bin"
+
+# Build dynamo wheel
+RUN source /opt/dynamo/venv/bin/activate && \
+    cd /workspace/lib/bindings/python && \
+    uv build --wheel --out-dir /workspace/dist --python 3.12 && \
+    if [ "$RELEASE_BUILD" = "true" ]; then \
+        uv build --wheel --out-dir /workspace/dist --python 3.11 && \
+        uv build --wheel --out-dir /workspace/dist --python 3.10; \
+    fi && \
+    cd /workspace && \
+    uv build --wheel --out-dir /workspace/dist && \
+    cd /workspace/deploy/dynamo/api-store && \
+    uv build --wheel --out-dir /workspace/dist
+
 #######################################
 ########## CI Minimum Image ###########
 #######################################
 FROM build AS ci_minimum
 
 COPY . /workspace
+COPY --from=wheel_builder /workspace/dist/ /workspace/dist/
+
+# Package the bindings
+RUN mkdir -p /opt/dynamo/bindings/wheels && \
+    mkdir /opt/dynamo/bindings/lib && \
+    cp dist/ai_dynamo*cp312*.whl /opt/dynamo/bindings/wheels/. && \
+    cp target/release/libdynamo_llm_capi.so /opt/dynamo/bindings/lib/. && \
+    cp -r lib/bindings/c/include /opt/dynamo/bindings/.
+
+RUN uv pip install /workspace/dist/ai_dynamo_runtime*cp312*.whl && \
+    uv pip install /workspace/dist/ai_dynamo*any.whl && \
+    uv pip install /workspace/dist/ai_dynamo_store*any.whl
 
 ##########################################
 ########## Perf Analyzer Image ###########
@@ -338,7 +382,7 @@ RUN mkdir bin &&  \
 ########################################
 ########## Development Image ###########
 ########################################
-FROM build AS dev
+FROM ci_minimum AS dev
 
 ARG GENAI_PERF_TAG
 
@@ -380,8 +424,8 @@ RUN apt-get update && \
     echo "source $VIRTUAL_ENV/bin/activate" >> ~/.bashrc
 
 # Install the wheels
-COPY --from=build /workspace/dist/*.whl wheelhouse/
-RUN uv pip install ai-dynamo[vllm] --find-links wheelhouse && \
+COPY --from=wheel_builder /workspace/dist/*.whl wheelhouse/
+RUN uv pip install ai-dynamo[vllm] --find-links wheelhouse  && \
     rm -r wheelhouse
 
 # Tell vllm to use the Dynamo LLM C API for KV Cache Routing

container/build.sh

@@ -170,6 +170,9 @@ get_options() {
                 missing_requirement $1
             fi
             ;;
+        --release-build)
+            RELEASE_BUILD=true
+            ;;
         --)
             shift
             break
@@ -340,6 +343,10 @@ fi
 if [ ! -z ${HF_TOKEN} ]; then
     BUILD_ARGS+=" --build-arg HF_TOKEN=${HF_TOKEN} "
 fi
+if [  ! -z ${RELEASE_BUILD} ]; then
+    echo "Performing a release build!"
+    BUILD_ARGS+=" --build-arg RELEASE_BUILD=${RELEASE_BUILD} "
+fi
 
 LATEST_TAG="--tag dynamo:latest-${FRAMEWORK,,}"
 if [ ! -z ${TARGET} ]; then