docker环境内UIE服务化部署模型失败，报错 The FastDeploy didn't compile with Paddle Inference.

[z5z56]

• 【FastDeploy版本】：fastdeploy:1.0.4-gpu-cuda11.4-trt8.5-21.10
• 【系统平台】: Linux x64(Ubuntu 18.04)
• 【硬件】： i9-13900K， Nvidia GPU 3090TI

按照
https://github.com/PaddlePaddle/FastDeploy/blob/develop/examples/text/uie/serving/README_CN.md
在fastdeploy:1.0.4-gpu-cuda11.4-trt8.5-21.10的docker环境内，服务化部署UIE模型，启动失败

输出：
root@xmdx:/# CUDA_VISIBLE_DEVICES=0 fastdeployserver --model-repository=/hdd/ljz/FastDeploy/examples/text/uie/serving/models/ --backend-config=python,shm-default-byte-size=10485760
I0320 08:03:03.270487 274 metrics.cc:298] Collecting metrics for GPU 0: NVIDIA GeForce RTX 3090 Ti
I0320 08:03:03.379255 274 pinned_memory_manager.cc:240] Pinned memory pool is created at '0x7f4ace000000' with size 268435456
I0320 08:03:03.379357 274 cuda_memory_manager.cc:105] CUDA memory pool is created on device 0 with size 67108864
I0320 08:03:03.379930 274 model_repository_manager.cc:1022] loading: uie:1
I0320 08:03:03.510236 274 python.cc:1875] TRITONBACKEND_ModelInstanceInitialize: uie_0 (GPU device 0)
model_config: {'name': 'uie', 'platform': '', 'backend': 'python', 'version_policy': {'latest': {'num_versions': 1}}, 'max_batch_size': 1, 'input': [{'name': 'INPUT_0', 'data_type': 'TYPE_STRING', 'format': 'FORMAT_NONE', 'dims': [1], 'is_shape_tensor': False, 'allow_ragged_batch': False}, {'name': 'INPUT_1', 'data_type': 'TYPE_STRING', 'format': 'FORMAT_NONE', 'dims': [1], 'is_shape_tensor': False, 'allow_ragged_batch': False}], 'output': [{'name': 'OUTPUT_0', 'data_type': 'TYPE_STRING', 'dims': [1], 'label_filename': '', 'is_shape_tensor': False}], 'batch_input': [], 'batch_output': [], 'optimization': {'priority': 'PRIORITY_DEFAULT', 'execution_accelerators': {'gpu_execution_accelerator': [{'name': 'paddle', 'parameters': {'cpu_threads': '12'}}], 'cpu_execution_accelerator': []}, 'input_pinned_memory': {'enable': True}, 'output_pinned_memory': {'enable': True}, 'gather_kernel_buffer_threshold': 0, 'eager_batching': False}, 'instance_group': [{'name': 'uie_0', 'kind': 'KIND_GPU', 'count': 1, 'gpus': [0], 'secondary_devices': [], 'profile': [], 'passive': False, 'host_policy': ''}], 'default_model_filename': '', 'cc_model_filenames': {}, 'metric_tags': {}, 'parameters': {}, 'model_warmup': []}
input: ['INPUT_0', 'INPUT_1']
output: ['OUTPUT_0']
[ERROR] fastdeploy/runtime/runtime_option.cc(133)::UsePaddleBackend The FastDeploy didn't compile with Paddle Inference.

提示The FastDeploy didn't compile with Paddle Inference.，但是我没有做修改，而且之前跑OCR的服务化部署跑通过，也是使用 Paddle Inference后端+paddle的模型，在UIE服务化就启动不起来，不知道是什么问题，求大佬们赐教

[LouisHeck]

我也遇到同样的问题

[z5z56]

hi，请问你解决了吗

[LouisHeck]

hi，请问你解决了吗

你好，暂时解决部分问题。只能用ORT后端推理，PaddleInference和TensorRT后端推理，还是无法用。

镜像版本：paddlepaddle/fastdeploy1.0.5-gpu-cuda11.4-trt8.5-21.10
修复命令：python3 -m pip install --upgrade --force-reinstall fastdeploy-gpu-python -f https://www.paddlepaddle.org.cn/whl/fastdeploy.html
解决思路：强制重新安装fastdeploy-gpu-python

解决后问题：只能用ORT后端推理，PaddleInference和TensorRT后端推理，还是无法用。

截图一：部署UIE，服务成功启动截图

截图二：采用TensorRT部署，只启动了ORT后端，没有启动TensorRT后端

截图三：修改采用Paddle推理，启动失败，并且触发服务自动重启

[blakeliu]

镜像paddlepaddle/fastdeploy:1.0.0-cpu-only-21.10, 在系统win10 and ubuntu18.04 遇到同样的问题, paddle, onnx, openvio均不支持！

[ERROR] fastdeploy/runtime.cc(262)::UsePaddleBackend    The FastDeploy didn't compile with Paddle Inference.
[ERROR] fastdeploy/runtime.cc(271)::UseOrtBackend       The FastDeploy didn't compile with OrtBackend.
[ERROR] fastdeploy/runtime.cc(296)::UseOpenVINOBackend  The FastDeploy didn't compile with OpenVINO.

ubuntu18.04环境如下：
os

Ubuntu 18.04.6 LTS \n \l
5.4.0-150-generic #167~18.04.1-Ubuntu SMP Wed May 24 00:51:42 UTC 2023 x86_64 x86_64 x86_64 GNU/Linux

docker

Docker version 20.10.17, build 100c701

win10环境如下：

11th Gen Intel(R) Core(TM) i5-1135G7 @ 2.40GHz   2.42 GHz
64 位操作系统, 基于 x64 的处理器
Windows 10 家庭中文版
版本号：22H2
操作系统内部版本：19045.3324

docker:

PS D:\uie\serving> docker --version
Docker version 24.0.5, build ced0996

[blakeliu]

如同issues 1077 更换为fastdeploy_python-1.0.7-cp38-cp38-manylinux1_x86_64.whl，依然报错！

[huangjun11]

请问问题有得到解决吗，我使用ORT、paddle、openvino都报错。

[neo502721]

指定一下fastdeploy-python的版本，跟镜像的fastdeploy保持一致

python3 -m pip uninstall fastdeploy-python
python3 -m pip install fastdeploy-python=1.0.4 -f https://www.paddlepaddle.org.cn/whl/fastdeploy.html

[neo502721]

指定一下fastdeploy-python的版本，跟镜像的fastdeploy保持一致

python3 -m pip uninstall fastdeploy-python
python3 -m pip install fastdeploy-python=1.0.4 -f https://www.paddlepaddle.org.cn/whl/fastdeploy.html

GPU镜像使用:

python3 -m pip uninstall fastdeploy-python
python3 -m pip install fastdeploy-gpu-python==1.0.4 -f https://www.paddlepaddle.org.cn/whl/fastdeploy.html

[DDUFlyme]

docker run -dit --name uie_base --gpus all -p 8201:8201 --shm-size="2g" -v `pwd`:/uie_serving paddlepaddle/fastdeploy:1.0.7-gpu-cuda11.6-trt8.5-22.12

python3 -m pip install fastdeploy-gpu-python==1.0.7 -f https://www.paddlepaddle.org.cn/whl/fastdeploy.html

CUDA_VISIBLE_DEVICES=0 fastdeployserver --model-repository=/uie_serving/models --backend-config=python,shm-default-byte-size=10485760

第三步命令启动后报错：fastdeployserver: error while loading shared libraries: libdcgm.so.2: cannot open shared object file: No such file or directory，但系统中存在 libdcgm.so.3，请大佬们赐教 👀 👀

[neoragex2002]

use this:

wget https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2004/x86_64/datacenter-gpu-manager_2.2.9_amd64.deb
dpkg -i datacenter-gpu-manager_2.2.9_amd64.deb