Skipping loop optimisation error in tensorflow_serving

General Discussion
, , ,
#1

Hi team,
I am getting the following message when testing a model with tensorflow_serving,
can u please help me debug and understand what it means

2022-02-13 23:20:52.460513: I tensorflow_serving/model_servers/server_core.cc:486] Finished adding/updating models
2022-02-13 23:20:52.460589: I tensorflow_serving/model_servers/server.cc:367] Profiler service is enabled
2022-02-13 23:20:52.461847: I tensorflow_serving/model_servers/server.cc:393] Running gRPC ModelServer at 0.0.0.0:8500 ...
[warn] getaddrinfo: address family for nodename not supported
[evhttp_server.cc : 245] NET_LOG: Entering the event loop ...
2022-02-13 23:20:52.463794: I tensorflow_serving/model_servers/server.cc:414] Exporting HTTP/REST API at:localhost:8501 ...
2022-02-13 23:21:14.903736: W external/org_tensorflow/tensorflow/core/grappler/optimizers/loop_optimizer.cc:906] Skipping loop optimization for Merge node with control input: StatefulPartitionedCall/StatefulPartitionedCall/Incompatible_Data_Type_6/AssertGuard/branch_executed/_539

thanks

1 Like
#3

@Prajwal, Skipping loop optimization for Merge node with control input is a warning raised by Tensorflow (tensorflow/core/grappler/optimizers/loop_optimizer.cc), which skips the optimization of graph control flow. This loop optimization is used to improve the performance of TensorFlow models. You can safely ignore this warning while serving models with Tensorflow Serving. Thank you!