tensorflow - TensorFlow 无法连接到 TensorFlow master。TPU 可能还没有准备好

Question

我正在关注 GCP 的 BERT TPU 教程。https://cloud.google.com/tpu/docs/tutorials/bert

但是，在训练步骤，我遇到以下错误，可以继续训练：

WARNING:tensorflow:Failed to connect to the Tensorflow master. The TPU worker may not be ready (still scheduling) or the Tensorflow master address is incorrect: got (grpc://10.240.1.2:8470).
W0404 03:06:16.688671 139878329321216 tpu_system_metadata.py:97] Failed to connect to the Tensorflow master. The TPU worker may not be ready (still scheduling) or the Tensorflow master address is incorrect: got (grpc://10.240.1.2:8470).
WARNING:tensorflow:Retrying (2/288).
W0404 03:06:16.688920 139878329321216 tpu_system_metadata.py:98] Retrying (2/288).
INFO:tensorflow:Querying Tensorflow master (grpc://10.240.1.2:8470) for TPU system metadata.
I0404 03:06:16.689001 139878329321216 tpu_system_metadata.py:78] Querying Tensorflow master (grpc://10.240.1.2:8470) for TPU system metadata.
2020-04-04 03:06:16.690159: W tensorflow/core/distributed_runtime/rpc/grpc_session.cc:370] GrpcSession::ListDevices will initialize the session with an empty graph and other defaults because the session has not yet been created.

有谁知道什么可能导致这个问题？谢谢！

标签： tensorflowgoogle-cloud-platformtpu