Serving: Running example/mnist_saved_model /tmp/mnist_model error _pywrap_tensorflow_internal.so: undefined symbol: PyClass_Type

[weizhenzhao]

Hi Dear
I met an error when running the example of the mnist_model of the tensorflow_serving:

root@weizhen-Lenovo-IdeaPad-Y470:/home/weizhen/serving# bazel-bin/tensorflow_serving/example/mnist_saved_model /tmp/mnist_model

with the version of
ubuntu :16
tensorflow :1.5.0
python : 3.5.2
If anything you can help . I will be much apperciated

And the details is as blow:

`

root@weizhen-Lenovo-IdeaPad-Y470:/home/weizhen/serving# bazel-bin/tensorflow_serving/example/mnist_saved_model /tmp/mnist_model
Traceback (most recent call last):
File "/home/weizhen/serving/bazel-bin/tensorflow_serving/example/mnist_saved_model.runfiles/org_tensorflow/tensorflow/python/pywrap_tensorflow.py", line 58, in
from tensorflow.python.pywrap_tensorflow_internal import *
File "/home/weizhen/serving/bazel-bin/tensorflow_serving/example/mnist_saved_model.runfiles/org_tensorflow/tensorflow/python/pywrap_tensorflow_internal.py", line 28, in
_pywrap_tensorflow_internal = swig_import_helper()
File "/home/weizhen/serving/bazel-bin/tensorflow_serving/example/mnist_saved_model.runfiles/org_tensorflow/tensorflow/python/pywrap_tensorflow_internal.py", line 24, in swig_import_helper
_mod = imp.load_module('_pywrap_tensorflow_internal', fp, pathname, description)
File "/usr/lib/python3.5/imp.py", line 242, in load_module
return load_dynamic(name, filename, file)
File "/usr/lib/python3.5/imp.py", line 342, in load_dynamic
return _load(spec)
ImportError: /home/weizhen/serving/bazel-bin/tensorflow_serving/example/mnist_saved_model.runfiles/org_tensorflow/tensorflow/python/_pywrap_tensorflow_internal.so: undefined symbol: PyClass_Type

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
File "/home/weizhen/serving/bazel-bin/tensorflow_serving/example/mnist_saved_model.runfiles/tf_serving/tensorflow_serving/example/mnist_saved_model.py", line 35, in
import tensorflow as tf
File "/home/weizhen/serving/bazel-bin/tensorflow_serving/example/mnist_saved_model.runfiles/org_tensorflow/tensorflow/__init__.py", line 24, in
from tensorflow.python import *
File "/home/weizhen/serving/bazel-bin/tensorflow_serving/example/mnist_saved_model.runfiles/org_tensorflow/tensorflow/python/__init__.py", line 49, in
from tensorflow.python import pywrap_tensorflow
File "/home/weizhen/serving/bazel-bin/tensorflow_serving/example/mnist_saved_model.runfiles/org_tensorflow/tensorflow/python/pywrap_tensorflow.py", line 74, in
raise ImportError(msg)
ImportError: Traceback (most recent call last):
File "/home/weizhen/serving/bazel-bin/tensorflow_serving/example/mnist_saved_model.runfiles/org_tensorflow/tensorflow/python/pywrap_tensorflow.py", line 58, in
from tensorflow.python.pywrap_tensorflow_internal import *
File "/home/weizhen/serving/bazel-bin/tensorflow_serving/example/mnist_saved_model.runfiles/org_tensorflow/tensorflow/python/pywrap_tensorflow_internal.py", line 28, in
_pywrap_tensorflow_internal = swig_import_helper()
File "/home/weizhen/serving/bazel-bin/tensorflow_serving/example/mnist_saved_model.runfiles/org_tensorflow/tensorflow/python/pywrap_tensorflow_internal.py", line 24, in swig_import_helper
_mod = imp.load_module('_pywrap_tensorflow_internal', fp, pathname, description)
File "/usr/lib/python3.5/imp.py", line 242, in load_module
return load_dynamic(name, filename, file)
File "/usr/lib/python3.5/imp.py", line 342, in load_dynamic
return _load(spec)
ImportError: /home/weizhen/serving/bazel-bin/tensorflow_serving/example/mnist_saved_model.runfiles/org_tensorflow/tensorflow/python/_pywrap_tensorflow_internal.so: undefined symbol: PyClass_Type

Failed to load the native TensorFlow runtime.

See https://www.tensorflow.org/install/install_sources#common_installation_problems

for some common reasons and solutions. Include the entire stack trace
above this error message when asking for help.
root@weizhen-Lenovo-IdeaPad-Y470:/home/weizhen/serving#
`

[weizhenzhao]

I found the problem that the tensorflow_serving I downloaded was python2.7 version
so in python 2.7 version .I didn't install tensorflow
But I think I used git clone --recurse-submodules https://github.com/tensorflow/serving to download
the tensorflow serving
So is the default version of tensorflow_serving that git donwloaded is python 2.7?

[NearLinHere]

I met the same problem under the same environment.
However, this error jumped out after "bazel run tensorflow/examples/speech_commands:generate_streaming_test_wav" this command.

[NearLinHere]

@weizhenzhao how did you fix it?

[weizhenzhao]

This problem is because that I compile the framework in python2.7 version and in python 2.7 there's no tensorflow installed . So change the python version to python 3.5 then the problem maybe can be solved

[weizhenzhao]

Or you can reference to the blob written by me : http://www.cnblogs.com/weizhen/p/8491236.html

[NearLinHere]

@weizhenzhao thanks for your sharing. May I ask that how to make ubuntu compile with python 3.5? I have no idea about this.

[robinqhuang]

@weizhenzhao @NearLinHere I built tensorflow r.16 and install it. The I built the latest serving from sources. I am using Python 3.5 virtual environment on ubuntu 16.04.

When I run bazel-bin/tensorflow_serving/example/mnist_saved_model /tmp/mnist_model. I had the same issue. It seems the source code of serving does not work with Python 3 after googling.

[elvys-zhang]

I have met the same problem. My python version is 3.5.2, and tensorflow serving version is 1.7.0, bazel version is 0.13.0. Then how could it be solved please?

[lenjoy]

probably due to package version. the good news is that the docker version is working, with r.18 https://www.tensorflow.org/serving/docker

[suenyandan]

I found the problem that the tensorflow_serving I downloaded was python2.7 version
so in python 2.7 version .I didn't install tensorflow
But I think I used git clone --recurse-submodules https://github.com/tensorflow/serving to download
the tensorflow serving
So is the default version of tensorflow_serving that git donwloaded is python 2.7?

Thanks for your sharing.I have run into the same error when runing "bazel test -c opt -- //tensorflow/... -//tensorflow/compiler/... -//tensorflow/contrib/lite/..." according to the official instructions to build tensorflow from source. So if the default version of tensorflow that git donwloaded is python 2.7, how can i download a repository with python version 3.x ?