tensorflow.python.framework.errors_impl.InvalidArgumentError: Unsuccessful TensorSliceReader constructor: Failed to get matching files on ./model/train_model.ckpt-36: Not found: ./model

[AndyShen105]

I find there are some people also has this problem. And most of them resolved this problem by add "./" before the path of logdir. But, it doesn't work for me. And the version of tensorflow in my computer is 1.3.0 .
Maybe there is someone who can help me solve this problem.Thanks!

sv = tf.train.Supervisor(is_chief=is_chief,
saver=saver,
global_step=global_step,
init_op=init_op,
logdir="./model", #this is the checkpoint_dir
checkpoint_basename="train_model.ckpt",
save_model_secs=10,
save_summaries_secs=10)
tf.train.Supervisor helps us save and restore model automatically, but it doesn't work.