DataTap Tensorflow Support
Tensorflow images support DataTaps by:
- Placing the shared library in the
/opt/bludata/
directory. - Installing and configuring the necessary Hadoop package.
- Configuring required environment variables.
After creating a virtual cluster with a Tensorflow image, log in to one of the virtual nodes/containers in that cluster, and then verify basic I/O functionality by executing the following commands in a Python shell:
import tensorflow as tf
import os
#check CXX11_ABI_FLAG
from tensorflow.python.framework.versionsimport CXX11_ABI_FLAG
CXX11_ABI_FLAG
#load bdfs shared library
bdfs_file_system_library= os.path.join("/opt/bluedata","libbdfs_file_system_shared_r1_13.so")
tf.load_library(bdfs_file_system_library)
#write to a test file
with tf.gfile.Open("dtap://TenantStorage/tensorflow/dtap.txt", 'w') as f:
f.write("This is the dtaptest file")
#read from the test file
with tf.gfile.Open("dtap://TenantStorage/tensorflow/dtap.txt", 'r') as f:
content = f.read()
# show the connect of the file
Content