相关疑难解决方法(0)

我有一个TensorFlow集群启动并运行,我正在尝试使用一个客户端进程将数据入队,并将其从另一个进程出列.我不能让这个工作,我做错了什么？

这是推送数据的程序:

# queue_push.py
import tensorflow as tf
import time

with tf.container("qtest"):
    q = tf.FIFOQueue(capacity=10, dtypes=[tf.float32],
                     shapes=[[]], name="q")
    v = tf.placeholder(tf.float32, shape=())
    enqueue = q.enqueue([v])

with tf.Session("grpc://localhost:2210") as sess:
    while True:
        t = time.time()
        print(t)
        sess.run(enqueue, feed_dict={v: t})
        time.sleep(1)

我的程序来拉数据:

# queue_pull.py
import tensorflow as tf
import time

with tf.container("qtest"):
    q = tf.FIFOQueue(capacity=10, dtypes=[tf.float32],
                     shapes=[[]], name="q")
    dequeue = q.dequeue()

with tf.Session("grpc://localhost:2222") as sess:
    while True:
        v = sess.run(dequeue)
        print("Pulled:", v)
        time.sleep(0.5)

当我运行它们时,这就是我得到的:

$ python queue_push.py
1472420887.974484
1472420888.991067
1472420889.995756
1472420890.998365
1472420892.001799 …