search

amazon-archives / amazon-dsstne

Deep Scalable Sparse Tensor Network Engine (DSSTNE) is an Amazon developed library for building Deep Learning (DL) machine learning (ML) models
Apache License 2.0
4.41k stars 731 forks source link

Problem with netcdf after built on container #234

Closed spacelover1 closed 4 years ago

spacelover1 commented 4 years ago

I'm kinda facing the same closed issue #25, except that I've built the engine on container, and after running this line in the example document generateNetCDF -d gl_input -i ml-20m_ratings -o gl_input.nc -f features_input -s samples_input -c, the similar error comes up generateNetCDF: command not found.

I got some errors for netcdf using this Dockerfile so I added these lines for installing it:

RUN wget http://archive.ubuntu.com/ubuntu/pool/universe/n/netcdf-cxx/netcdf-cxx_4.3.0+ds.orig.tar.gz && \
    tar xzf netcdf-cxx_4.3.0+ds.orig.tar.gz && \
    cd netcdf-cxx4-4.3.0 && \
    ./configure --disable-filter-testing && \
    make && \
    make install

Any suggestions to solve this? Regards,

spacelover1 commented 4 years ago

So I could finally build the engine successfully, I just wanted to post the changes I've made in Dockerfile that lead to the successful build:

Instead of RUN apt-get install -y libnetcdf-c++4-dev I put:

RUN wget http://archive.ubuntu.com/ubuntu/pool/universe/n/netcdf-cxx/netcdf-cxx_4.3.0+ds.orig.tar.gz && \
    tar xzf netcdf-cxx_4.3.0+ds.orig.tar.gz && \
    cd netcdf-cxx4-4.3.0 && \
    ./configure --disable-filter-testing && \
    make && \
    make install

and instead of ENV PATH=/opt/amazon/dsstne/bin/:${PATH} I used ENV PATH=/opt/amazon/dsstne/build/bin/:${PATH}

I also used Ubuntu 16.04 docker image. Hope this would be helpful.