Merge branch 'development' into 'master'

Development

See merge request !22

Dockerfile

@@ -15,7 +15,7 @@ RUN dnf -y upgrade && dnf -y install pkg-config libgomp libpng python38 && dnf c
     echo "/usr/local/nvidia/lib64" >> /etc/ld.so.conf.d/nvidia.conf
 
 #ENV
-ENV PATH /usr/local/nvidia/bin:/usr/local/cuda/bin:${PATH}
+ENV PATH /usr/local/nvidia/bin:/usr/local/cuda/bin:/opt/tritonserver/bin:${PATH}
 ENV LD_LIBRARY_PATH /usr/local/nvidia/lib:/usr/local/nvidia/lib64:/opt/tritonserver/lib
 
 #Link
@@ -34,10 +34,15 @@ COPY ["${TARBALL1}", "/"]
 COPY ["${TARBALLN}","/opt/"]
 
 WORKDIR /
-RUN tar -zxvhf ./${TARBALL1} && rm -fv ./${TARBALL1} && tar -xvhf ./opt/nvidia.rpms.tar -C ./opt && rm -fv ./opt/nvidia.rpms.tar \
-  && rpm -i /opt/baserpms/re2-20190801-1.el8.x86_64.rpm && rpm --nodeps -i /opt/nvidia.rpms/*.rpm \
-  && pip3 install -r opt/basewhls/requirements.txt --no-index --find-links /opt/basewhls && rm -r /opt/nvidia.rpms \
-  && rm -r opt/basewhls && rm -r /opt/baserpms && mkdir /models
+RUN tar -zxvhf ./${TARBALL1} && rm -fv ./${TARBALL1}  \
+  && rpm -i /opt/baserpms/re2-20190801-1.el8.x86_64.rpm \
+  && pip3 install -r opt/basewhls/requirements.txt --no-index --find-links /opt/basewhls \
+  && rm -r opt/basewhls && rm -r /opt/baserpms && mkdir /models && chown -R triton /opt/tritonserver
+
+RUN tar -xvhf ./opt/nvidia.rpms.tar -C ./opt && rm -fv ./opt/nvidia.rpms.tar \
+  && rpm --nodeps -i /opt/nvidia.rpms/*.rpm \
+  && rm -r /opt/nvidia.rpms
+
 
 ENV NCCL_VERSION 2.8.4
 ENV LIBRARY_PATH /usr/local/cuda/lib64/stubs
@@ -49,6 +54,6 @@ EXPOSE 8000
 EXPOSE 8001
 EXPOSE 8002
 #
-ENTRYPOINT ["/opt/tritonserver/bin/tritonserver", "--strict-model-config=false", "--model-repository", "/models"]
+ENTRYPOINT ["/opt/tritonserver/nvidia_entrypoint.sh"]
 
 HEALTHCHECK --timeout=30s CMD curl -fs http://localhost:8000/v2 || exit 1
\ No newline at end of file

hardening_manifest.yaml

@@ -42,7 +42,7 @@ resources:
     filename: "structure.tar.gz"
     validation:
       type: "sha256"
-      value: "229416ebf2c8b7f4a534b2ee70d38c878a66cc80554d022d8a0a38b872f870fa"
+      value: "34c1d8aaedb45c26c63b883bc87d506533ac120287ad3de9719c59295937a0a8"
 
 maintainers:
   - name: "Daniel Mumpton"

scripts/nvidia_entrypoint.sh

+#!/bin/bash
+# Copyright (c) 2019-2020, NVIDIA CORPORATION. All rights reserved.
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions
+# are met:
+#  * Redistributions of source code must retain the above copyright
+#    notice, this list of conditions and the following disclaimer.
+#  * Redistributions in binary form must reproduce the above copyright
+#    notice, this list of conditions and the following disclaimer in the
+#    documentation and/or other materials provided with the distribution.
+#  * Neither the name of NVIDIA CORPORATION nor the names of its
+#    contributors may be used to endorse or promote products derived
+#    from this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS ``AS IS'' AND ANY
+# EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+# PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE COPYRIGHT OWNER OR
+# CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+# EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+# PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+# PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY
+# OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+set -e
+cat <<EOF
+
+=============================
+== Triton Inference Server ==
+=============================
+
+NVIDIA Release ${NVIDIA_TRITON_SERVER_VERSION} (build ${NVIDIA_BUILD_ID})
+
+Copyright (c) 2018-2020, NVIDIA CORPORATION.  All rights reserved.
+
+Various files include modifications (c) NVIDIA CORPORATION.  All rights reserved.
+NVIDIA modifications are covered by the license terms that apply to the underlying
+project or file.
+EOF
+
+if [[ "$(find -L /usr -name libcuda.so.1 | grep -v "compat") " == " " || "$(ls /dev/nvidiactl 2>/dev/null) " == " " ]]; then
+  echo
+  echo "WARNING: The NVIDIA Driver was not detected.  GPU functionality will not be available."
+  echo "   Use 'nvidia-docker run' to start this container; see"
+  echo "   https://github.com/NVIDIA/nvidia-docker/wiki/nvidia-docker ."
+  ln -s `find / -name libnvidia-ml.so -print -quit` /opt/tritonserver/lib/libnvidia-ml.so.1
+  export LD_LIBRARY_PATH=/usr/local/cuda-11.2/compat:${LD_LIBRARY_PATH}
+  export TRITON_SERVER_CPU_ONLY=1
+else
+  ( /usr/local/bin/checkSMVER.sh )
+  DRIVER_VERSION=$(sed -n 's/^NVRM.*Kernel Module *\([0-9.]*\).*$/\1/p' /proc/driver/nvidia/version 2>/dev/null || true)
+  if [[ ! "$DRIVER_VERSION" =~ ^[0-9]*.[0-9]*(.[0-9]*)?$ ]]; then
+    echo "Failed to detect NVIDIA driver version."
+  elif [[ "${DRIVER_VERSION%%.*}" -lt "${CUDA_DRIVER_VERSION%%.*}" ]]; then
+    if [[ "${_CUDA_COMPAT_STATUS}" == "CUDA Driver OK" ]]; then
+      echo
+      echo "NOTE: Legacy NVIDIA Driver detected.  Compatibility mode ENABLED."
+    else
+      echo
+      echo "ERROR: This container was built for NVIDIA Driver Release ${CUDA_DRIVER_VERSION%.*} or later, but"
+      echo "       version ${DRIVER_VERSION} was detected and compatibility mode is UNAVAILABLE."
+      echo
+      echo "       [[${_CUDA_COMPAT_STATUS}]]"
+      sleep 2
+    fi
+  fi
+fi
+
+if ! cat /proc/cpuinfo | grep flags | sort -u | grep avx >& /dev/null; then
+  echo
+  echo "ERROR: This container was built for CPUs supporting at least the AVX instruction set, but"
+  echo "       the CPU detected was $(cat /proc/cpuinfo |grep "model name" | sed 's/^.*: //' | sort -u), which does not report"
+  echo "       support for AVX.  An Illegal Instrution exception at runtime is likely to result."
+  echo "       See https://en.wikipedia.org/wiki/Advanced_Vector_Extensions#CPUs_with_AVX ."
+  sleep 2
+fi
+
+if [[ "$(df -k /dev/shm |grep ^shm |awk '{print $2}') " == "65536 " ]]; then
+  echo
+  echo "NOTE: The SHMEM allocation limit is set to the default of 64MB.  This may be"
+  echo "   insufficient for the inference server.  NVIDIA recommends the use of the following flags:"
+  echo "   nvidia-docker run --shm-size=1g --ulimit memlock=-1 --ulimit stack=67108864 ..."
+fi
+
+echo
+
+if [[ $# -eq 0 ]]; then
+  exec "/bin/bash"
+else
+  exec "$@"
+fi