Skip to content

Commit 11852d7

Browse files
Test sm without oss compliance and with efa
1 parent c4c8185 commit 11852d7

File tree

1 file changed

+33
-32
lines changed

1 file changed

+33
-32
lines changed

vllm/x86_64/gpu/Dockerfile

Lines changed: 33 additions & 32 deletions
Original file line numberDiff line numberDiff line change
@@ -1,6 +1,7 @@
11
FROM docker.io/vllm/vllm-openai:v0.11.0 as base
22
ARG PYTHON="python3"
33
LABEL maintainer="Amazon AI"
4+
ARG EFA_VERSION="1.43.3"
45
LABEL dlc_major_version="1"
56
ENV DEBIAN_FRONTEND=noninteractive \
67
LANG=C.UTF-8 \
@@ -10,51 +11,43 @@ ENV DEBIAN_FRONTEND=noninteractive \
1011
# Force stdin, stdout and stderr to be totally unbuffered. Good for logging
1112
PYTHONDONTWRITEBYTECODE=1 \
1213
PYTHONUNBUFFERED=1 \
13-
PYTHONIOENCODING=UTF-8
14+
PYTHONIOENCODING=UTF-8 \
15+
LD_LIBRARY_PATH="/usr/local/lib:/opt/amazon/ofi-nccl/lib/x86_64-linux-gnu:/opt/amazon/openmpi/lib:/opt/amazon/efa/lib:/usr/local/cuda/lib64:${LD_LIBRARY_PATH}" \
16+
PATH="/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda/bin:${PATH}"
1417

1518
WORKDIR /
1619

1720
COPY deep_learning_container.py /usr/local/bin/deep_learning_container.py
1821
COPY bash_telemetry.sh /usr/local/bin/bash_telemetry.sh
22+
COPY install_efa.sh install_efa.sh
1923

2024
RUN chmod +x /usr/local/bin/deep_learning_container.py && \
2125
chmod +x /usr/local/bin/bash_telemetry.sh && \
22-
echo 'source /usr/local/bin/bash_telemetry.sh' >> /etc/bash.bashrc && \
26+
echo 'source /usr/local/bin/bash_telemetry.sh' >> /etc/bash.bashrc
2327
# OSS compliance and software update
24-
apt-get update && \
25-
apt-get upgrade -y && \
26-
apt-get install -y --allow-change-held-packages --no-install-recommends unzip && \
27-
apt-get clean && \
28-
HOME_DIR=/root && \
29-
curl -o ${HOME_DIR}/oss_compliance.zip https://aws-dlinfra-utilities.s3.amazonaws.com/oss_compliance.zip && \
30-
unzip ${HOME_DIR}/oss_compliance.zip -d ${HOME_DIR}/ && \
31-
cp ${HOME_DIR}/oss_compliance/test/testOSSCompliance /usr/local/bin/testOSSCompliance && \
32-
chmod +x /usr/local/bin/testOSSCompliance && \
33-
chmod +x ${HOME_DIR}/oss_compliance/generate_oss_compliance.sh && \
34-
${HOME_DIR}/oss_compliance/generate_oss_compliance.sh ${HOME_DIR} ${PYTHON} && \
35-
# create symlink for python
36-
ln -s /usr/bin/python3 /usr/bin/python && \
37-
# clean up
38-
rm -rf ${HOME_DIR}/oss_compliance* && \
39-
rm -rf /tmp/tmp* && \
40-
rm -rf /tmp/uv* && \
41-
rm -rf /var/lib/apt/lists/* && \
42-
rm -rf /root/.cache | true
43-
44-
# ====================== ec2 =========================================
45-
FROM base AS vllm-ec2
28+
# apt-get update && \
29+
# apt-get upgrade -y && \
30+
# apt-get install -y --allow-change-held-packages --no-install-recommends unzip && \
31+
# apt-get clean && \
32+
# HOME_DIR=/root && \
33+
# curl -o ${HOME_DIR}/oss_compliance.zip https://aws-dlinfra-utilities.s3.amazonaws.com/oss_compliance.zip && \
34+
# unzip ${HOME_DIR}/oss_compliance.zip -d ${HOME_DIR}/ && \
35+
# cp ${HOME_DIR}/oss_compliance/test/testOSSCompliance /usr/local/bin/testOSSCompliance && \
36+
# chmod +x /usr/local/bin/testOSSCompliance && \
37+
# chmod +x ${HOME_DIR}/oss_compliance/generate_oss_compliance.sh && \
38+
# ${HOME_DIR}/oss_compliance/generate_oss_compliance.sh ${HOME_DIR} ${PYTHON} && \
39+
# # create symlink for python
40+
# ln -s /usr/bin/python3 /usr/bin/python && \
41+
# # clean up
42+
# rm -rf ${HOME_DIR}/oss_compliance* && \
43+
# rm -rf /tmp/tmp* && \
44+
# rm -rf /tmp/uv* && \
45+
# rm -rf /var/lib/apt/lists/* && \
46+
# rm -rf /root/.cache | true
4647

47-
ARG EFA_VERSION="1.43.3"
48-
ENV LD_LIBRARY_PATH="/usr/local/lib:/opt/amazon/ofi-nccl/lib/x86_64-linux-gnu:/opt/amazon/openmpi/lib:/opt/amazon/efa/lib:/usr/local/cuda/lib64:${LD_LIBRARY_PATH}" \
49-
PATH="/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda/bin:${PATH}"
50-
51-
# Install EFA
52-
COPY install_efa.sh install_efa.sh
53-
COPY dockerd_entrypoint.sh /usr/local/bin/dockerd_entrypoint.sh
5448

5549
RUN bash install_efa.sh ${EFA_VERSION} && \
5650
rm install_efa.sh && \
57-
chmod +x /usr/local/bin/dockerd_entrypoint.sh && \
5851
mkdir -p /tmp/nvjpeg \
5952
&& cd /tmp/nvjpeg \
6053
&& wget https://developer.download.nvidia.com/compute/cuda/redist/libnvjpeg/linux-x86_64/libnvjpeg-linux-x86_64-12.4.0.76-archive.tar.xz \
@@ -68,6 +61,14 @@ RUN bash install_efa.sh ${EFA_VERSION} && \
6861
&& rm -rf /usr/local/cuda/bin/cuobjdump* \
6962
&& rm -rf /usr/local/cuda/bin/nvdisasm*
7063

64+
65+
# ====================== ec2 =========================================
66+
FROM base AS vllm-ec2
67+
68+
# Install EFA
69+
COPY dockerd_entrypoint.sh /usr/local/bin/dockerd_entrypoint.sh
70+
RUN chmod +x /usr/local/bin/dockerd_entrypoint.sh
71+
7172
ENTRYPOINT ["/usr/local/bin/dockerd_entrypoint.sh"]
7273
# ====================== sagemaker =========================================
7374
FROM base AS vllm-sagemaker

0 commit comments

Comments
 (0)