如何解决 tf_serving_entrypoint.sh: line 3: 6 Illegal instruction (core dumped) when using tensorflow/serving image

Question

我在使用 docker 图像 tesorflow/serving:1.13.0 在云中部署我的模型时遇到了这个问题。但它在我的本地系统中运行完美。

云系统实际日志为：

usr/bin/tf_serving_entrypoint.sh: line 3:     6 Illegal instruction     (core dumped) tensorflow_model_server --port=8500 --rest_api_port=8501 --model_name=${MODEL_NAME} --model_base_path=${MODEL_BASE_PATH}/${MODEL_NAME} "$@"

我尝试使用来自 tensorflow 服务 docker 注册表的各种图像，其中 none 有效。

这是我的 docker-compose 文件结构。以及挂载文件结构。 Structure of mounting folder

  tensorflow:
image: tensorflow/serving:1.13.0
container_name: tensorflow
environment:
  - MODEL_NAME=test
volumes:
  - ./data_pipeline/machine_learning/models/v1/:/models/test/1
ports:
  - 8501:8501

我希望得到以下结果，这样我们就可以使用为结果提供服务的模型。这些是我在容器为运行.

时获得的日志

2019-05-08 06:31:31.357564: I tensorflow_serving/model_servers/server.cc:82] Building single TensorFlow model file config:  model_name: test model_base_path: /models/test

2019-05-08 06:31:31.388148: I tensorflow_serving/model_servers/server_core.cc:461] Adding/updating models.

2019-05-08 06:31:31.388179: I tensorflow_serving/model_servers/server_core.cc:558]  (Re-)adding model: test

2019-05-08 06:31:31.496616: I tensorflow_serving/core/basic_manager.cc:739] Successfully reserved resources to load servable {name: test version: 1}

2019-05-08 06:31:31.496640: I tensorflow_serving/core/loader_harness.cc:66] Approving load for servable version {name: test version: 1}

2019-05-08 06:31:31.496651: I tensorflow_serving/core/loader_harness.cc:74] Loading servable version {name: test version: 1}

2019-05-08 06:31:31.496663: I external/org_tensorflow/tensorflow/contrib/session_bundle/bundle_shim.cc:363] Attempting to load native SavedModelBundle in bundle-shim from: /models/test/1

2019-05-08 06:31:31.496669: I external/org_tensorflow/tensorflow/cc/saved_model/reader.cc:31] Reading SavedModel from: /models/test/1

2019-05-08 06:31:31.600082: I external/org_tensorflow/tensorflow/cc/saved_model/reader.cc:54] Reading meta graph with tags { serve }

2019-05-08 06:31:31.626460: I external/org_tensorflow/tensorflow/core/platform/cpu_feature_guard.cc:141] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA

2019-05-08 06:31:31.657342: I external/org_tensorflow/tensorflow/cc/saved_model/loader.cc:182] Restoring SavedModel bundle.

2019-05-08 06:31:31.863963: I external/org_tensorflow/tensorflow/cc/saved_model/loader.cc:285] SavedModel load for tags { serve }; Status: success. Took 367280 microseconds.

2019-05-08 06:31:31.864020: I tensorflow_serving/servables/tensorflow/saved_model_warmup.cc:101] No warmup data file found at /models/test/1/assets.extra/tf_serving_warmup_requests

2019-05-08 06:31:31.864115: I tensorflow_serving/core/loader_harness.cc:86] Successfully loaded servable version {name: test version: 1}

2019-05-08 06:31:31.875615: I tensorflow_serving/model_servers/server.cc:313] Running gRPC ModelServer at 0.0.0.0:8500 ...

[warn] getaddrinfo: address family for nodename not supported
2019-05-08 06:31:31.883332: I tensorflow_serving/model_servers/server.cc:333] Exporting HTTP/REST API at:localhost:8501 ...

[evhttp_server.cc : 237] RAW: Entering the event loop ...

有人可以帮我解决这个问题吗？

Answer 1

我已经通过为我正在处理的 CPU 构建二进制文件解决了这个错误。

我已经从这个 link 构建了二进制文件。 tensorflow-serving from source using docker

我已将图像推送到 dockerhub 存储库。如果有人不想使用与我的 CPU 相同的配置构建自己的相应图像。

Dockerhub repository for tensorflow-serving images for Centos built from source

如何解决 tf_serving_entrypoint.sh: line 3: 6 Illegal instruction (core dumped) when using tensorflow/serving image

How to solve tf_serving_entrypoint.sh: line 3: 6 Illegal instruction (core dumped) when using tensorflow/serving image

python

machine-learning

docker

tensorflow-serving