Error in building from src, got 'undeclared inclusion(s) in rule' in mpi.so generation

[liismn]

I want to know the right bazel version for this project. Any suggestion will be appreciated.

Environment info

Operating System: CentOS 7.2

Installed version of CUDA and cuDNN: none

If installed from source, provide

1. The commit hash (git rev-parse HEAD)
2. The output of bazel version

[root@localhost tensorflow-allreduce]# git rev-parse HEAD
44302f0961e76002989523c2f424a011e54fd806
[root@localhost tensorflow-allreduce]# bazel version
............
Build label: 0.4.3
Build target: bazel-out/local-fastbuild/bin/src/main/java/com/google/devtools/build/lib/bazel/BazelServer_deploy.jar
Build time: Thu Dec 22 12:31:25 2016 (1482409885)
Build timestamp: 1482409885
Build timestamp as int: 1482409885
[root@localhost mnist]# gcc --version
gcc (GCC) 4.8.5 20150623 (Red Hat 4.8.5-11)
Copyright (C) 2015 Free Software Foundation, Inc.
This is free software; see the source for copying conditions.  There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.

If possible, provide a minimal reproducible example (We usually don't have time to read hundreds of lines of your code)

After configure success...
[root@localhost tensorflow-allreduce]# bazel build --config=opt //tensorflow/tools/pip_package:build_pip_package

What other attempted solutions have you tried?

Logs or other output that would be helpful

(If logs are large, please upload as attachment or provide link).

 #  warning _FORTIFY_SOURCE requires compiling with optimization (-O)
    ^
ERROR: /root/download/tensorflow-allreduce/tensorflow/contrib/mpi/BUILD:13:1: undeclared inclusion(s) in rule '//tensorflow/contrib/mpi:mpi.so':
this rule is missing dependency declarations for the following files included by 'tensorflow/contrib/mpi/mpi_ops.cc':
  '/root/download/tensorflow-allreduce/tensorflow/stream_executor/lib/statusor.h'
  '/root/download/tensorflow-allreduce/tensorflow/stream_executor/platform/port.h'
  '/root/download/tensorflow-allreduce/tensorflow/stream_executor/lib/error.h'
  '/root/download/tensorflow-allreduce/tensorflow/stream_executor/lib/status.h'
  '/root/download/tensorflow-allreduce/tensorflow/stream_executor/platform/logging.h'.
In file included from /usr/lib/gcc/x86_64-redhat-linux/4.8.5/../../../../include/c++/4.8.5/x86_64-redhat-linux/bits/os_defines.h:39:0,
                 from /usr/lib/gcc/x86_64-redhat-linux/4.8.5/../../../../include/c++/4.8.5/x86_64-redhat-linux/bits/c++config.h:2097,
                 from /usr/lib/gcc/x86_64-redhat-linux/4.8.5/../../../../include/c++/4.8.5/utility:68,
                 from /usr/lib/gcc/x86_64-redhat-linux/4.8.5/../../../../include/c++/4.8.5/unordered_map:38,
                 from tensorflow/contrib/mpi/mpi_ops.cc:16:
/usr/include/features.h:330:4: warning: #warning _FORTIFY_SOURCE requires compiling with optimization (-O) [-Wcpp]
 #  warning _FORTIFY_SOURCE requires compiling with optimization (-O)
    ^
tensorflow/contrib/mpi/mpi_ops.cc: In instantiation of 'void tensorflow::contrib::mpi::MPIAllgatherOp<Device>::ComputeAsync(tensorflow::OpKernelContext*, tensorflow::AsyncOpKernel::DoneCallback) [with Device = Eigen::ThreadPoolDevice; tensorflow::AsyncOpKernel::DoneCallback = std::function<void()>]':
tensorflow/contrib/mpi/mpi_ops.cc:1147:1:   required from here
tensorflow/contrib/mpi/mpi_ops.cc:1086:12: warning: unused variable 'device_context' [-Wunused-variable]
       auto device_context = context->op_device_context();
            ^
tensorflow/contrib/mpi/mpi_ops.cc: In instantiation of 'void tensorflow::contrib::mpi::MPIAllreduceOp<Device>::ComputeAsync(tensorflow::OpKernelContext*, tensorflow::AsyncOpKernel::DoneCallback) [with Device = Eigen::ThreadPoolDevice; tensorflow::AsyncOpKernel::DoneCallback = std::function<void()>]':
tensorflow/contrib/mpi/mpi_ops.cc:1147:1:   required from here
tensorflow/contrib/mpi/mpi_ops.cc:1018:12: warning: unused variable 'device_context' [-Wunused-variable]
       auto device_context = context->op_device_context();
            ^
Target //tensorflow/tools/pip_package:build_pip_package failed to build

[aburden5]

Hi liismn - did you ever solve this issue? I am having the exact same problem, and haven't had any success tracking down a solution.

[liismn]

No. I tried with bazel 0.4.4 and 0.4.5, and got similar errors.

[aburden5]

I can get past that error by adding this path to the deps in tensorflow/contrib/mpi/BUILD:
"//tensorflow/stream_executor"
and removing "//tensorflow/core:lib" from the disallowed_deps for tf_custom_op_library in tensorflow/tensorflow.bzl

can now build tf, but when I try "import tensorflow.contrib.mpi" I get "WARNING:tensorflow:mpi.so file could not be loaded."

[liismn]

"WARNING:tensorflow:mpi.so file could not be loaded."

You can try to add the_path_to_mpi_lib to LD_LIBRARY_PATH

[huansong]

Same error here. Did anyone have a solution to that yet? Also, this issue seems to be only associated when using CPU-only configuration.