tensorflow缺少mpi_rendezvous_mgr的依赖项声明

Question

我正在尝试在SL7.3上编译tensorflow，支持动词和MPI (没有cuda)。在尝试使用bazel构建时，我遇到了以下错误：

用于编译的命令：

bazel build --config=opt //tensorflow/tools/pip_package:build_pip_package --verbose_failures

看到的错误：

ERROR: /root/tensorflow/tensorflow/contrib/mpi/BUILD:60:1: undeclared inclusion(s) in rule '//tensorflow/contrib/mpi:mpi_rendezvous_mgr':
this rule is missing dependency declarations for the following files included by 'tensorflow/contrib/mpi/mpi_rendezvous_mgr.cc':
  '/root/tensorflow/tensorflow/core/distributed_runtime/tensor_coding.h'.
Target //tensorflow/tools/pip_package:build_pip_package failed to build

有没有人能帮我一下？我试图在stackoverflow (和其他论坛)上寻找解决方案，但找不到太多信息。

谢谢

Answer 1

您可以尝试在.../tensorflow/tensorflow/contrib/mpi/BUILD文件中添加一行

...

cc_library(
    name = "mpi_rendezvous_mgr",
    srcs = ["mpi_rendezvous_mgr.cc"],
    hdrs = ["mpi_rendezvous_mgr.h"],
    deps = [
        ":mpi_msg_proto_cc",
        ":mpi_utils",
        "//tensorflow/core:core_cpu_internal",
        "//tensorflow/core:framework",
        "//tensorflow/core:gpu_runtime",
        "//tensorflow/core:lib",
        "//tensorflow/core:protos_cc",
        "//tensorflow/core:worker_proto_cc",
        "//tensorflow/core/distributed_runtime:base_rendezvous_mgr",
        "//tensorflow/core/distributed_runtime:session_mgr",
        "//tensorflow/core/distributed_runtime:worker_env",
        "//third_party/mpi",
        "//tensorflow/core/distributed_runtime:tensor_coding",   ## <<<<< add this line
    ],
)

希望能有所帮助，

BR/thupalo