Feature: nvBLAS and OpenBLAS plugin (#539)

* add openblas plugin, update gpu docker images with netlib-lgpl

* update images and plugins

* add nvblas plugin

* revert gpu docker image change, add -Pnetlib-lgpl to base images

* change configuraitons to functions, add pugins to cluster.yaml

aztk/models/plugins/internal/plugin_manager.py

@@ -23,6 +23,8 @@ class PluginManager:
         hdfs=plugins.HDFSPlugin,
         simple=plugins.SimplePlugin,
         spark_ui_proxy=plugins.SparkUIProxyPlugin,
+        openblas=plugins.OpenBLASPlugin,
+        nvblas=plugins.NvBLASPlugin,
     )
 
     def __init__(self):

aztk/spark/models/plugins/__init__.py

@@ -5,3 +5,5 @@ from .resource_monitor import ResourceMonitorPlugin
 from .rstudio_server import RStudioServerPlugin
 from .simple import SimplePlugin
 from .spark_ui_proxy import SparkUIProxyPlugin
+from .openblas import OpenBLASPlugin
+from .nvblas import NvBLASPlugin

aztk/spark/models/plugins/nvblas/__init__.py

@@ -0,0 +1 @@
+from .configuration import *

aztk/spark/models/plugins/nvblas/configuration.py

@@ -0,0 +1,18 @@
+import os
+from aztk.models.plugins.plugin_configuration import PluginConfiguration, PluginPort, PluginTargetRole
+from aztk.models.plugins.plugin_file import PluginFile
+from aztk.utils import constants
+
+dir_path = os.path.dirname(os.path.realpath(__file__))
+
+
+def NvBLASPlugin():
+    return PluginConfiguration(
+        name="nvblas",
+        ports=[],
+        target_role=PluginTargetRole.All,
+        execute="nvblas.sh",
+        files=[
+            PluginFile("nvblas.sh", os.path.join(dir_path, "nvblas.sh")),
+        ]
+    )

aztk/spark/models/plugins/nvblas/nvblas.sh

@@ -0,0 +1,65 @@
+#!/bin/bash
+
+apt-get update &&
+apt-get install -y libblas-dev liblapack-dev &&
+update-alternatives --config libblas.so.3
+update-alternatives --config liblapack.so.3
+
+export NVBLAS_CONFIG_FILE=/usr/local/cuda/lib64/nvblas.conf
+echo "export NVBLAS_CONFIG_FILE=/usr/local/cuda/lib64/nvblas.conf" >> ~/.bashrc
+
+echo '# This is the configuration file to use NVBLAS Library
+# Setup the environment variable NVBLAS_CONFIG_FILE to specify your own config file.
+# By default, if NVBLAS_CONFIG_FILE is not defined, 
+# NVBLAS Library will try to open the file "nvblas.conf" in its current directory
+# Example : NVBLAS_CONFIG_FILE  /home/cuda_user/my_nvblas.conf
+# The config file should have restricted write permissions accesses
+
+# Specify which output log file (default is stderr)
+NVBLAS_LOGFILE  /root/nvblas.log
+
+# Enable trace log of every intercepted BLAS calls
+NVBLAS_TRACE_LOG_ENABLED
+
+#Put here the CPU BLAS fallback Library of your choice
+#It is strongly advised to use full path to describe the location of the CPU Library
+NVBLAS_CPU_BLAS_LIB  /usr/lib/libblas.so
+
+# List of GPU devices Id to participate to the computation 
+# Use ALL if you want all your GPUs to contribute
+# Use ALL0, if you want all your GPUs of the same type as device 0 to contribute
+# However, NVBLAS consider that all GPU have the same performance and PCI bandwidth
+# By default if no GPU are listed, only device 0 will be used
+
+#NVBLAS_GPU_LIST 0 2 4
+#NVBLAS_GPU_LIST ALL
+NVBLAS_GPU_LIST ALL0
+
+# Tile Dimension
+NVBLAS_TILE_DIM 2048
+
+# Autopin Memory
+NVBLAS_AUTOPIN_MEM_ENABLED
+
+#List of BLAS routines that are prevented from running on GPU (use for debugging purpose
+# The current list of BLAS routines supported by NVBLAS are
+# GEMM, SYRK, HERK, TRSM, TRMM, SYMM, HEMM, SYR2K, HER2K
+
+#NVBLAS_GPU_DISABLED_SGEMM 
+#NVBLAS_GPU_DISABLED_DGEMM 
+#NVBLAS_GPU_DISABLED_CGEMM 
+#NVBLAS_GPU_DISABLED_ZGEMM 
+
+# Computation can be optionally hybridized between CPU and GPU
+# By default, GPU-supported BLAS routines are ran fully on GPU
+# The option NVBLAS_CPU_RATIO_<BLAS_ROUTINE> give the ratio [0,1] 
+# of the amount of computation that should be done on CPU
+# CAUTION : this option should be used wisely because it can actually
+# significantly reduced the overall performance if too much work is given to CPU
+
+#NVBLAS_CPU_RATIO_CGEMM 0.07' > $NVBLAS_CONFIG_FILE
+
+export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/lib/libblas:/usr/local/cuda/lib64
+echo "export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/lib/libblas:/usr/local/cuda/lib64" >> ~/.bashrc
+export LD_PRELOAD=/usr/local/cuda/lib64/libnvblas.so
+echo "export LD_PRELOAD=/usr/local/cuda/lib64/libnvblas.so" >> ~/.bashrc

aztk/spark/models/plugins/openblas/__init__.py

@@ -0,0 +1 @@
+from .configuration import *

aztk/spark/models/plugins/openblas/configuration.py

@@ -0,0 +1,18 @@
+import os
+from aztk.models.plugins.plugin_configuration import PluginConfiguration, PluginPort, PluginTargetRole
+from aztk.models.plugins.plugin_file import PluginFile
+from aztk.utils import constants
+
+dir_path = os.path.dirname(os.path.realpath(__file__))
+
+
+def OpenBLASPlugin():
+    return PluginConfiguration(
+        name="openblas",
+        ports=[],
+        target_role=PluginTargetRole.All,
+        execute="openblas.sh",
+        files=[
+            PluginFile("openblas.sh", os.path.join(dir_path, "openblas.sh")),
+        ],
+    )

aztk/spark/models/plugins/openblas/openblas.sh

@@ -0,0 +1,4 @@
+#!/bin/bash
+apt-get update &&
+apt-get install -y libopenblas-base &&
+update-alternatives --config libblas.so.3

aztk_cli/config/cluster.yaml

@@ -5,7 +5,7 @@
 # Toolkit configuration [Required] You can use `aztk toolkit` command to find which are the available tookits
 toolkit:
   software: spark
-  version: 2.2.0
+  version: 2.3.0
   # Which environemnt is needed for spark anaconda, r, miniconda
   environment: {environment}
   # Optional version for the environment
@@ -16,7 +16,7 @@ toolkit:
 
 
 # vm_size: <vm-size, see available options here: https://azure.microsoft.com/en-us/pricing/details/virtual-machines/linux/>
-vm_size: standard_a2
+vm_size: standard_f2
 
 # size: <number of dedicated nodes in the cluster, not that clusters must contain all dedicated or all low priority nodes>
 size: 2
@@ -39,11 +39,13 @@ username: spark
 
 # Enable plugins
 plugins:
-  # - name: spark_ui_proxy
   # - name: jupyterlab
   # - name: jupyter
   # - name: hdfs
   # - name: rstudio_server
+  # - name: spark_ui_proxy
+  # - name: openblas
+  # - name: nvblas
 
 # Allow master node to also be a worker <true/false> (Default: true)
 # worker_on_master: true

docker-image/base/spark1.6.3/Dockerfile

@@ -58,7 +58,7 @@ RUN apt-get clean \
     && cd spark \
     && git checkout tags/v${SPARK_VERSION_KEY} \
     && export MAVEN_OPTS="-Xmx3g -XX:ReservedCodeCacheSize=1024m" \
-    && ./make-distribution.sh --name custom-spark --tgz -Phive -Phive-thriftserver -Dhadoop.version=${HADOOP_VERSION} -Phadoop-2.6 -DskipTests \
+    && ./make-distribution.sh --name custom-spark --tgz -Pnetlib-lgpl -Phive -Phive-thriftserver -Dhadoop.version=${HADOOP_VERSION} -Phadoop-2.6 -DskipTests \
     && tar -xvzf /spark/spark-${SPARK_VERSION_KEY}-bin-custom-spark.tgz --directory=/home \
     && ln -s "/home/spark-${SPARK_VERSION_KEY}-bin-custom-spark" /home/spark-current \
     && rm -rf /spark \ 

docker-image/base/spark2.1.0/Dockerfile

@@ -58,7 +58,7 @@ RUN apt-get clean \
     && cd spark \
     && git checkout tags/v${SPARK_VERSION_KEY} \
     && export MAVEN_OPTS="-Xmx3g -XX:ReservedCodeCacheSize=1024m" \
-    && ./dev/make-distribution.sh --name custom-spark --pip --tgz -Phive -Phive-thriftserver -Dhadoop.version=${HADOOP_VERSION} -DskipTests \
+    && ./dev/make-distribution.sh --name custom-spark --pip --tgz -Pnetlib-lgpl -Phive -Phive-thriftserver -Dhadoop.version=${HADOOP_VERSION} -DskipTests \
     && tar -xvzf /spark/spark-${SPARK_VERSION_KEY}-bin-custom-spark.tgz --directory=/home \
     && ln -s "/home/spark-${SPARK_VERSION_KEY}-bin-custom-spark" /home/spark-current \
     && rm -rf /spark \ 

docker-image/base/spark2.2.0/Dockerfile

@@ -57,7 +57,7 @@ RUN apt-get clean \
     && cd spark \
     && git checkout tags/v${SPARK_VERSION_KEY} \
     && export MAVEN_OPTS="-Xmx3g -XX:ReservedCodeCacheSize=1024m" \
-    && ./dev/make-distribution.sh --name custom-spark --pip --tgz -Phive -Phive-thriftserver -Dhadoop.version=${HADOOP_VERSION} -DskipTests \
+    && ./dev/make-distribution.sh --name custom-spark --pip --tgz -Pnetlib-lgpl -Phive -Phive-thriftserver -Dhadoop.version=${HADOOP_VERSION} -DskipTests \
     && tar -xvzf /spark/spark-${SPARK_VERSION_KEY}-bin-custom-spark.tgz --directory=/home \
     && ln -s "/home/spark-${SPARK_VERSION_KEY}-bin-custom-spark" /home/spark-current \
     && rm -rf /spark \ 

docker-image/base/spark2.3.0/Dockerfile

@@ -58,7 +58,7 @@ RUN apt-get clean \
     && cd spark \
     && git checkout tags/v${SPARK_VERSION_KEY} \
     && export MAVEN_OPTS="-Xmx3g -XX:ReservedCodeCacheSize=1024m" \
-    && ./dev/make-distribution.sh --name custom-spark --pip --tgz -Phive -Phive-thriftserver -Dhadoop.version=${HADOOP_VERSION} -DskipTests \
+    && ./dev/make-distribution.sh --name custom-spark --pip --tgz -Pnetlib-lgpl -Phive -Phive-thriftserver -Dhadoop.version=${HADOOP_VERSION} -DskipTests \
     && tar -xvzf /spark/spark-${SPARK_VERSION_KEY}-bin-custom-spark.tgz --directory=/home \
     && ln -s "/home/spark-${SPARK_VERSION_KEY}-bin-custom-spark" /home/spark-current \
     && rm -rf /spark \ 

docker-image/gpu/spark1.6.3/Dockerfile

@@ -2,12 +2,14 @@ FROM  aztk/spark:v0.1.0-spark1.6.3-base
 
 LABEL com.nvidia.volumes.needed="nvidia_driver"
 
-RUN NVIDIA_GPGKEY_SUM=d1be581509378368edeec8c1eb2958702feedf3bc3d17011adbf24efacce4ab5 && \
+RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates apt-transport-https gnupg-curl && \
+    rm -rf /var/lib/apt/lists/* && \
+    NVIDIA_GPGKEY_SUM=d1be581509378368edeec8c1eb2958702feedf3bc3d17011adbf24efacce4ab5 && \
     NVIDIA_GPGKEY_FPR=ae09fe4bbd223a84b2ccfce3f60f4b3d7fa2af80 && \
-    apt-key adv --fetch-keys http://developer.download.nvidia.com/compute/cuda/repos/ubuntu1604/x86_64/7fa2af80.pub && \
+    apt-key adv --fetch-keys https://developer.download.nvidia.com/compute/cuda/repos/ubuntu1604/x86_64/7fa2af80.pub && \
     apt-key adv --export --no-emit-version -a $NVIDIA_GPGKEY_FPR | tail -n +5 > cudasign.pub && \
     echo "$NVIDIA_GPGKEY_SUM  cudasign.pub" | sha256sum -c --strict - && rm cudasign.pub && \
-    echo "deb http://developer.download.nvidia.com/compute/cuda/repos/ubuntu1604/x86_64 /" > /etc/apt/sources.list.d/cuda.list
+    echo "deb https://developer.download.nvidia.com/compute/cuda/repos/ubuntu1604/x86_64 /" > /etc/apt/sources.list.d/cuda.list
 
 # CUDA
 ENV CUDA_VERSION 8.0.61
@@ -45,9 +47,13 @@ ENV LD_LIBRARY_PATH /usr/local/nvidia/lib:/usr/local/nvidia/lib64
 
 ENV LIBRARY_PATH /usr/local/cuda/lib64/stubs:${LIBRARY_PATH}
 
+# nvidia-container-runtime
+ENV NVIDIA_VISIBLE_DEVICES all
+ENV NVIDIA_DRIVER_CAPABILITIES compute,utility
+ENV NVIDIA_REQUIRE_CUDA "cuda>=8.0"
 
 # cuDNN
-RUN echo "deb http://developer.download.nvidia.com/compute/machine-learning/repos/ubuntu1604/x86_64 /" > /etc/apt/sources.list.d/nvidia-ml.list
+RUN echo "deb https://developer.download.nvidia.com/compute/machine-learning/repos/ubuntu1604/x86_64 /" > /etc/apt/sources.list.d/nvidia-ml.list
 
 ENV CUDNN_VERSION 6.0.21
 
@@ -72,8 +78,4 @@ ENV NUMBAPRO_LIBDEVICE /usr/local/cuda/nvvm/libdevice/
 ENV NUMBAPRO_NVVM /usr/local/cuda-8.0/nvvm/lib64/libnvvm.so
 ENV NUMBAPRO_CUDALIB /usr/local/cuda-8.0/targets/x86_64-linux/lib/
 
-# # Tensorflow
-# RUN pip install --upgrade tensorflow-gpu
-
-WORKDIR $SPARK_HOME
-CMD ["bin/spark-class", "org.apache.spark.deploy.master.Master"]
+CMD ["/bin/bash"]

docker-image/gpu/spark2.1.0/Dockerfile

@@ -2,12 +2,14 @@ FROM  aztk/spark:v0.1.0-spark2.1.0-base
 
 LABEL com.nvidia.volumes.needed="nvidia_driver"
 
-RUN NVIDIA_GPGKEY_SUM=d1be581509378368edeec8c1eb2958702feedf3bc3d17011adbf24efacce4ab5 && \
+RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates apt-transport-https gnupg-curl && \
+    rm -rf /var/lib/apt/lists/* && \
+    NVIDIA_GPGKEY_SUM=d1be581509378368edeec8c1eb2958702feedf3bc3d17011adbf24efacce4ab5 && \
     NVIDIA_GPGKEY_FPR=ae09fe4bbd223a84b2ccfce3f60f4b3d7fa2af80 && \
-    apt-key adv --fetch-keys http://developer.download.nvidia.com/compute/cuda/repos/ubuntu1604/x86_64/7fa2af80.pub && \
+    apt-key adv --fetch-keys https://developer.download.nvidia.com/compute/cuda/repos/ubuntu1604/x86_64/7fa2af80.pub && \
     apt-key adv --export --no-emit-version -a $NVIDIA_GPGKEY_FPR | tail -n +5 > cudasign.pub && \
     echo "$NVIDIA_GPGKEY_SUM  cudasign.pub" | sha256sum -c --strict - && rm cudasign.pub && \
-    echo "deb http://developer.download.nvidia.com/compute/cuda/repos/ubuntu1604/x86_64 /" > /etc/apt/sources.list.d/cuda.list
+    echo "deb https://developer.download.nvidia.com/compute/cuda/repos/ubuntu1604/x86_64 /" > /etc/apt/sources.list.d/cuda.list
 
 # CUDA
 ENV CUDA_VERSION 8.0.61
@@ -45,9 +47,13 @@ ENV LD_LIBRARY_PATH /usr/local/nvidia/lib:/usr/local/nvidia/lib64
 
 ENV LIBRARY_PATH /usr/local/cuda/lib64/stubs:${LIBRARY_PATH}
 
+# nvidia-container-runtime
+ENV NVIDIA_VISIBLE_DEVICES all
+ENV NVIDIA_DRIVER_CAPABILITIES compute,utility
+ENV NVIDIA_REQUIRE_CUDA "cuda>=8.0"
 
 # cuDNN
-RUN echo "deb http://developer.download.nvidia.com/compute/machine-learning/repos/ubuntu1604/x86_64 /" > /etc/apt/sources.list.d/nvidia-ml.list
+RUN echo "deb https://developer.download.nvidia.com/compute/machine-learning/repos/ubuntu1604/x86_64 /" > /etc/apt/sources.list.d/nvidia-ml.list
 
 ENV CUDNN_VERSION 6.0.21
 
@@ -72,8 +78,4 @@ ENV NUMBAPRO_LIBDEVICE /usr/local/cuda/nvvm/libdevice/
 ENV NUMBAPRO_NVVM /usr/local/cuda-8.0/nvvm/lib64/libnvvm.so
 ENV NUMBAPRO_CUDALIB /usr/local/cuda-8.0/targets/x86_64-linux/lib/
 
-# # Tensorflow
-# RUN pip install --upgrade tensorflow-gpu
-
-WORKDIR $SPARK_HOME
-CMD ["bin/spark-class", "org.apache.spark.deploy.master.Master"]
+CMD ["/bin/bash"]

docker-image/gpu/spark2.2.0/Dockerfile

@@ -2,12 +2,14 @@ FROM  aztk/spark:v0.1.0-spark2.2.0-base
 
 LABEL com.nvidia.volumes.needed="nvidia_driver"
 
-RUN NVIDIA_GPGKEY_SUM=d1be581509378368edeec8c1eb2958702feedf3bc3d17011adbf24efacce4ab5 && \
+RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates apt-transport-https gnupg-curl && \
+    rm -rf /var/lib/apt/lists/* && \
+    NVIDIA_GPGKEY_SUM=d1be581509378368edeec8c1eb2958702feedf3bc3d17011adbf24efacce4ab5 && \
     NVIDIA_GPGKEY_FPR=ae09fe4bbd223a84b2ccfce3f60f4b3d7fa2af80 && \
-    apt-key adv --fetch-keys http://developer.download.nvidia.com/compute/cuda/repos/ubuntu1604/x86_64/7fa2af80.pub && \
+    apt-key adv --fetch-keys https://developer.download.nvidia.com/compute/cuda/repos/ubuntu1604/x86_64/7fa2af80.pub && \
     apt-key adv --export --no-emit-version -a $NVIDIA_GPGKEY_FPR | tail -n +5 > cudasign.pub && \
     echo "$NVIDIA_GPGKEY_SUM  cudasign.pub" | sha256sum -c --strict - && rm cudasign.pub && \
-    echo "deb http://developer.download.nvidia.com/compute/cuda/repos/ubuntu1604/x86_64 /" > /etc/apt/sources.list.d/cuda.list
+    echo "deb https://developer.download.nvidia.com/compute/cuda/repos/ubuntu1604/x86_64 /" > /etc/apt/sources.list.d/cuda.list
 
 # CUDA
 ENV CUDA_VERSION 8.0.61
@@ -45,9 +47,13 @@ ENV LD_LIBRARY_PATH /usr/local/nvidia/lib:/usr/local/nvidia/lib64
 
 ENV LIBRARY_PATH /usr/local/cuda/lib64/stubs:${LIBRARY_PATH}
 
+# nvidia-container-runtime
+ENV NVIDIA_VISIBLE_DEVICES all
+ENV NVIDIA_DRIVER_CAPABILITIES compute,utility
+ENV NVIDIA_REQUIRE_CUDA "cuda>=8.0"
 
 # cuDNN
-RUN echo "deb http://developer.download.nvidia.com/compute/machine-learning/repos/ubuntu1604/x86_64 /" > /etc/apt/sources.list.d/nvidia-ml.list
+RUN echo "deb https://developer.download.nvidia.com/compute/machine-learning/repos/ubuntu1604/x86_64 /" > /etc/apt/sources.list.d/nvidia-ml.list
 
 ENV CUDNN_VERSION 6.0.21
 
@@ -72,8 +78,4 @@ ENV NUMBAPRO_LIBDEVICE /usr/local/cuda/nvvm/libdevice/
 ENV NUMBAPRO_NVVM /usr/local/cuda-8.0/nvvm/lib64/libnvvm.so
 ENV NUMBAPRO_CUDALIB /usr/local/cuda-8.0/targets/x86_64-linux/lib/
 
-# # Tensorflow
-# RUN pip install --upgrade tensorflow-gpu
-
-WORKDIR $SPARK_HOME
-CMD ["bin/spark-class", "org.apache.spark.deploy.master.Master"]
+CMD ["/bin/bash"]

docker-image/gpu/spark2.3.0/Dockerfile

@@ -2,12 +2,14 @@ FROM  aztk/spark:v0.1.0-spark2.3.0-base
 
 LABEL com.nvidia.volumes.needed="nvidia_driver"
 
-RUN NVIDIA_GPGKEY_SUM=d1be581509378368edeec8c1eb2958702feedf3bc3d17011adbf24efacce4ab5 && \
+RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates apt-transport-https gnupg-curl && \
+    rm -rf /var/lib/apt/lists/* && \
+    NVIDIA_GPGKEY_SUM=d1be581509378368edeec8c1eb2958702feedf3bc3d17011adbf24efacce4ab5 && \
     NVIDIA_GPGKEY_FPR=ae09fe4bbd223a84b2ccfce3f60f4b3d7fa2af80 && \
-    apt-key adv --fetch-keys http://developer.download.nvidia.com/compute/cuda/repos/ubuntu1604/x86_64/7fa2af80.pub && \
+    apt-key adv --fetch-keys https://developer.download.nvidia.com/compute/cuda/repos/ubuntu1604/x86_64/7fa2af80.pub && \
     apt-key adv --export --no-emit-version -a $NVIDIA_GPGKEY_FPR | tail -n +5 > cudasign.pub && \
     echo "$NVIDIA_GPGKEY_SUM  cudasign.pub" | sha256sum -c --strict - && rm cudasign.pub && \
-    echo "deb http://developer.download.nvidia.com/compute/cuda/repos/ubuntu1604/x86_64 /" > /etc/apt/sources.list.d/cuda.list
+    echo "deb https://developer.download.nvidia.com/compute/cuda/repos/ubuntu1604/x86_64 /" > /etc/apt/sources.list.d/cuda.list
 
 # CUDA
 ENV CUDA_VERSION 8.0.61
@@ -45,9 +47,13 @@ ENV LD_LIBRARY_PATH /usr/local/nvidia/lib:/usr/local/nvidia/lib64
 
 ENV LIBRARY_PATH /usr/local/cuda/lib64/stubs:${LIBRARY_PATH}
 
+# nvidia-container-runtime
+ENV NVIDIA_VISIBLE_DEVICES all
+ENV NVIDIA_DRIVER_CAPABILITIES compute,utility
+ENV NVIDIA_REQUIRE_CUDA "cuda>=8.0"
 
 # cuDNN
-RUN echo "deb http://developer.download.nvidia.com/compute/machine-learning/repos/ubuntu1604/x86_64 /" > /etc/apt/sources.list.d/nvidia-ml.list
+RUN echo "deb https://developer.download.nvidia.com/compute/machine-learning/repos/ubuntu1604/x86_64 /" > /etc/apt/sources.list.d/nvidia-ml.list
 
 ENV CUDNN_VERSION 6.0.21
 
@@ -72,8 +78,4 @@ ENV NUMBAPRO_LIBDEVICE /usr/local/cuda/nvvm/libdevice/
 ENV NUMBAPRO_NVVM /usr/local/cuda-8.0/nvvm/lib64/libnvvm.so
 ENV NUMBAPRO_CUDALIB /usr/local/cuda-8.0/targets/x86_64-linux/lib/
 
-# # Tensorflow
-# RUN pip install --upgrade tensorflow-gpu
-
-WORKDIR $SPARK_HOME
-CMD ["bin/spark-class", "org.apache.spark.deploy.master.Master"]
+CMD ["/bin/bash"]