amazon-web-services - Tensorflow Serving服务器未启动
问题描述
当我使用 TF-serving 从 S3(本地 minio)部署 tensorflow 模型时,似乎可以找到该模型,但服务器没有启动。作为比较,我还在从 S3 下载后部署了模型,它工作正常:
下载模型,安装并服务:<--这很好用
docker run -t --rm -p 8501:8501 \
-v "$TESTDATA/monkey-species:/models/monkey-species" \
-e MODEL_NAME=monkey-species \
tensorflow/serving
2021-04-06 13:37:04.648259: I tensorflow_serving/model_servers/server.cc:88] Building single TensorFlow model file config: model_name: monkey-species model_base_path: /models/monkey-species
2021-04-06 13:37:04.648455: I tensorflow_serving/model_servers/server_core.cc:464] Adding/updating models.
2021-04-06 13:37:04.648490: I tensorflow_serving/model_servers/server_core.cc:587] (Re-)adding model: monkey-species
2021-04-06 13:37:04.760432: I tensorflow_serving/core/basic_manager.cc:740] Successfully reserved resources to load servable {name: monkey-species version: 1}
2021-04-06 13:37:04.760496: I tensorflow_serving/core/loader_harness.cc:66] Approving load for servable version {name: monkey-species version: 1}
2021-04-06 13:37:04.760519: I tensorflow_serving/core/loader_harness.cc:74] Loading servable version {name: monkey-species version: 1}
2021-04-06 13:37:04.761104: I external/org_tensorflow/tensorflow/cc/saved_model/reader.cc:32] Reading SavedModel from: /models/monkey-species/1
2021-04-06 13:37:04.895710: I external/org_tensorflow/tensorflow/cc/saved_model/reader.cc:55] Reading meta graph with tags { serve }
2021-04-06 13:37:04.895776: I external/org_tensorflow/tensorflow/cc/saved_model/reader.cc:93] Reading SavedModel debug info (if present) from: /models/monkey-species/1
2021-04-06 13:37:04.896701: I external/org_tensorflow/tensorflow/core/platform/cpu_feature_guard.cc:142] This TensorFlow binary is optimized with oneAPI Deep Neural Network Library (oneDNN) to use the following CPU instructions in performance-critical operations: AVX2 FMA
To enable them in other operations, rebuild TensorFlow with the appropriate compiler flags.
2021-04-06 13:37:05.179687: I external/org_tensorflow/tensorflow/cc/saved_model/loader.cc:206] Restoring SavedModel bundle.
2021-04-06 13:37:05.222351: I external/org_tensorflow/tensorflow/core/platform/profile_utils/cpu_utils.cc:112] CPU Frequency: 2400000000 Hz
2021-04-06 13:37:06.276588: I external/org_tensorflow/tensorflow/cc/saved_model/loader.cc:190] Running initialization op on SavedModel bundle at path: /models/monkey-species/1
2021-04-06 13:37:06.489424: I external/org_tensorflow/tensorflow/cc/saved_model/loader.cc:277] SavedModel load for tags { serve }; Status: success: OK. Took 1728328 microseconds.
2021-04-06 13:37:06.554367: I tensorflow_serving/servables/tensorflow/saved_model_warmup_util.cc:59] No warmup data file found at /models/monkey-species/1/assets.extra/tf_serving_warmup_requests
2021-04-06 13:37:06.563250: I tensorflow_serving/core/loader_harness.cc:87] Successfully loaded servable version {name: monkey-species version: 1}
2021-04-06 13:37:06.567457: I tensorflow_serving/model_servers/server.cc:371] Running gRPC ModelServer at 0.0.0.0:8500 ...
[warn] getaddrinfo: address family for nodename not supported
[evhttp_server.cc : 238] NET_LOG: Entering the event loop ...
2021-04-06 13:37:06.569568: I tensorflow_serving/model_servers/server.cc:391] Exporting HTTP/REST API at:localhost:8501 ...
但是,如果我直接从 Minio S3 使用模型,它会在打印 CPU 频率后停止:<--这不起作用
docker run -t --rm -p 8501:8501 \
-e AWS_ACCESS_KEY_ID=$AWS_ID \
-e AWS_SECRET_ACCESS_KEY=$AWS_KEY \
-e S3_USE_HTTPS=0 \
-e S3_VERIFY_SSL=0 \
-e S3_ENDPOINT=$ENDPOINT \
tensorflow/serving --model_name=monkey-model --model_base_path=s3://monkey-species/monkey-model --monitoring_config_file=/var/config/monitoring_config.txt &
2021-04-06 13:41:11.779649: I tensorflow_serving/model_servers/server.cc:88] Building single TensorFlow model file config: model_name: monkey-model model_base_path: s3://monkey-species/monkey-model
2021-04-06 13:41:11.779823: I tensorflow_serving/model_servers/server_core.cc:464] Adding/updating models.
2021-04-06 13:41:11.779890: I tensorflow_serving/model_servers/server_core.cc:587] (Re-)adding model: monkey-model
2021-04-06 13:41:17.450543: I tensorflow_serving/core/basic_manager.cc:740] Successfully reserved resources to load servable {name: monkey-model version: 1}
2021-04-06 13:41:17.450631: I tensorflow_serving/core/loader_harness.cc:66] Approving load for servable version {name: monkey-model version: 1}
2021-04-06 13:41:17.450707: I tensorflow_serving/core/loader_harness.cc:74] Loading servable version {name: monkey-model version: 1}
2021-04-06 13:41:17.874780: I external/org_tensorflow/tensorflow/cc/saved_model/reader.cc:32] Reading SavedModel from: s3://monkey-species/monkey-model/1
2021-04-06 13:41:42.407789: I external/org_tensorflow/tensorflow/cc/saved_model/reader.cc:55] Reading meta graph with tags { serve }
2021-04-06 13:41:42.407851: I external/org_tensorflow/tensorflow/cc/saved_model/reader.cc:93] Reading SavedModel debug info (if present) from: s3://monkey-species/monkey-model/1
2021-04-06 13:41:42.646294: I external/org_tensorflow/tensorflow/core/platform/cpu_feature_guard.cc:142] This TensorFlow binary is optimized with oneAPI Deep Neural Network Library (oneDNN) to use the following CPU instructions in performance-critical operations: AVX2 FMA
To enable them in other operations, rebuild TensorFlow with the appropriate compiler flags.
2021-04-06 13:41:42.932037: I external/org_tensorflow/tensorflow/cc/saved_model/loader.cc:206] Restoring SavedModel bundle.
2021-04-06 13:41:43.396767: I external/org_tensorflow/tensorflow/core/platform/profile_utils/cpu_utils.cc:112] CPU Frequency: 2400000000 Hz
我等了大约 15 分钟,但在那之后什么也没发生,这意味着模式服务器不可用于推理……有人知道吗?
解决方案
推荐阅读
- javascript - 动态下拉菜单 - 根据使用 JavaScript 选择的选项更改按钮 URL
- python - 在运行 Spark 群集的 Azure Databricks 中需要 Azure Blob 存储
- javascript - PM2 在使用 watch 选项重新启动时返回错误“spawn ps”
- android - 当应用中使用的 APK 需要较高的最低 sdk 时,如何使用最低 sdk 来支持较旧的 android 软件?
- reactjs - 无法使用 webrtc MediaDevices 在反应应用程序中切换摄像头(从前到后)
- css - 如何将 CSS 文件的 CSS 规则包含到另一个 CSS 文件的某些选择器中?
- .net-core - Dotnet Core 如何获取 JWT 标头部分?
- angularjs - 在 agularjs 的 ng-options 中选择的选项
- mysql - 删除 SQL 重复项
- c# - 如何在 c# 中使用 AddYears 方法获得 2 月 29 日