CODEWITHSUNDEEP
compilationredhatglibctensorflowbazel
read Read
read Read

Reputation: 6063

How to install bazel and tensorflow on Red Hat 6.7

I would like to install bazel from source, and use bazel to compile tensorflow on a cluster running redhat 6.7. When I try to install bazel, the glibc version (2.12) is too old. I do not have root access to the cluster. Is it possible to install tensorflow in this case?

My system information:

-bash-4.1$ cat /etc/redhat-release
Red Hat Enterprise Linux Server release 6.7 (Santiago)
-bash-4.1$ which gcc
/usr/bin/gcc
-bash-4.1$ gcc -v
Using built-in specs.
Target: x86_64-redhat-linux
Configured with: ../configure --prefix=/usr --mandir=/usr/share/man --infodir=/usr/share/info --with-bugurl=http://bugzilla.redhat.com/bugzilla --enable-bootstrap --enable-shared --enable-threads=posix --enable-checking=release --with-system-zlib --enable-__cxa_atexit --disable-libunwind-exceptions --enable-gnu-unique-object --enable-languages=c,c++,objc,obj-c++,java,fortran,ada --enable-java-awt=gtk --disable-dssi --with-java-home=/usr/lib/jvm/java-1.5.0-gcj-1.5.0.0/jre --enable-libgcj-multifile --enable-java-maintainer-mode --with-ecj-jar=/usr/share/java/eclipse-ecj.jar --disable-libjava-multilib --with-ppl --with-cloog --with-tune=generic --with-arch_32=i686 --build=x86_64-redhat-linux
Thread model: posix
gcc version 4.4.7 20120313 (Red Hat 4.4.7-16) (GCC) 
-bash-4.1$ ldd --version
ldd (GNU libc) 2.12

The system has newer gcc installed as well. I tried using it, bazel still won't compile.

-bash-4.1$ /usr/local/gcc/4.8.4/bin/gcc -v
Using built-in specs.
COLLECT_GCC=/usr/local/gcc/4.8.4/bin/gcc
COLLECT_LTO_WRAPPER=/usr/local/gcc/4.8.4/libexec/gcc/x86_64-unknown-linux-gnu/4.8.4/lto-wrapper
Target: x86_64-unknown-linux-gnu
Configured with: ../configure --prefix=/usr/local/gcc/4.8.4
Thread model: posix
gcc version 4.8.4 (GCC)

When I was compiling bazel, I got the following error:

bazel-0.1.1/_bin/build-runfiles: /usr/lib64/libstdc++.so.6: version `GLIBCXX_3.4.14' not found

Some people also reported this issue: https://github.com/tensorflow/tensorflow/issues/110 and https://github.com/tensorflow/tensorflow/issues/527

How can I install the missing dependency locally, and have bazel pick up the right library?

Upvotes: 4

Views: 6844

Answers (2)

kftse
kftse

Reputation: 191

In case anyone need to do this manually:

  1. Compile up-to-date glibc, gcc and all their dependencies from source with the option --disable-rpath to avoid glibc path being hard coded to system default. Do NOT add glibc to LD_LIBRARY_PATH directly, or all executables including rm will stop working.

  2. Compile python with your gcc, install pip and the official wheel

./configure --prefix=$PWD/build --enable-unicode=ucs4 --with-cxx-main=g++ && make && make install

  1. Start python with the following command to load the correct glibc, where ${GLIBC_PATH} is your install location to glibc, make an alias

alias tensorflow='${GLIBC_PATH}/lib/ld-2.23.so --library-path ${GLIBC_PATH}/lib:${LD_LIBRARY_PATH}which python'

  1. Import tensorflow to check no error occur

After able to import tensorflow module without any error, you can use any other computer (possibly ubuntu VM on your PC) to compile a custom wheel with machine specific option, and copy that to your cluster following the guide

bazel build -c opt --copt=-mavx --copt=-mavx2 --copt=-mfma --copt=-mfpmath=both --copt=-msse4.2 --copt=-msse4.1 //tensorflow/tools/pip_package:build_pip_package

Upvotes: 1

Damien Martin-Guillerez
Damien Martin-Guillerez

Reputation: 2370

You should be able to compile from source with a newer version of Bazel exporting LD_FLAGS, CXX and CC, and tweaking the tools/cpp/CROSSTOOL file of Bazel. Please open a github issue on Bazel (https://github.com/bazelbuild/bazel/issues) if you have further questions.

I am currently working on making all that easier. Sorry for the mess.

Upvotes: 3

Related Questions