paddleocr gpu部署报错OSError: (External) CUDA error(3), initialization error. [Hint: 'cudaErrorInitializationError'. The API call failed because the CUDA driver and runtime could not be initialized. ] (at /paddle/paddle/fluid/platform/gpu_info.cc:355)

[sangwentao]

请提供下述完整信息以便快速定位问题/Please provide the following information to quickly locate the problem

• 系统环境/System Environment：
  阿里云服务器，阿里云内核
• 版本号/Version：Paddle： PaddleOCR： 问题相关组件/Related components：
  paddleocr2.3，uwsgi，nginx1.8.1，gpu T4 16G
• 运行指令/Command Code：
  uwsgi启动：uwsgi --ini uwsgi.ini
  [uwsgi] master=true socket = 127.0.0.1:8000 module = run:app chdir = /home/yunshuo/PaddleOCR-release-2.3/ wsgi-file = /home/yunshuo/PaddleOCR-release-2.3/run.py callable = app processes=1 threads=1 stats = 127.0.0.1:9191 pidfile=uwsgi.pid touch-reload=/home/yunshuo/PaddleOCR-release-2.3 daemonize = /home/logs/flask/server.log
  之后启动nginx，
  worker_processes 1;
• 完整报错/Complete Error Message：
  *** Starting uWSGI 2.0.19.1 (64bit) on [Tue Jan 25 14:07:16 2022] ***
  compiled with version: 9.3.0 on 02 March 2021 12:41:33
  os: Linux-5.10.60-9.al8.x86_64 Upload PaddleOCR code  #1 SMP Mon Sep 6 21:30:37 CST 2021
  nodename: xybot-gpu-001
  machine: x86_64
  clock source: unix
  pcre jit disabled
  detected number of CPU cores: 8
  current working directory: /home/yunshuo/PaddleOCR-release-2.3
  writing pidfile to uwsgi.pid
  detected binary path: /root/anaconda3/envs/ocr_python36/bin/uwsgi
  uWSGI running as root, you can use --uid/--gid/--chroot options
  *** WARNING: you are running uWSGI as root !!! (use the --uid flag) ***
  chdir() to /home/yunshuo/PaddleOCR-release-2.3/
  your processes number limit is 123187
  your memory page size is 4096 bytes
  detected max file descriptor number: 65535
  lock engine: pthread robust mutexes
  thunder lock: disabled (you can enable it with --thunder-lock)
  uwsgi socket 0 bound to TCP address 127.0.0.1:8000 fd 3
  uWSGI running as root, you can use --uid/--gid/--chroot options
  *** WARNING: you are running uWSGI as root !!! (use the --uid flag) ***
  Python version: 3.6.13 | packaged by conda-forge | (default, Feb 19 2021, 05:36:01) [GCC 9.3.0]
  Python main interpreter initialized at 0x563dadc964e0
  uWSGI running as root, you can use --uid/--gid/--chroot options
  *** WARNING: you are running uWSGI as root !!! (use the --uid flag) ***
  python threads support enabled
  your server socket listen backlog is limited to 100 connections
  your mercy for graceful operations on workers is 60 seconds
  mapped 145840 bytes (142 KB) for 1 cores
  *** Operational MODE: single process ***
  WSGI app 0 (mountpoint='') ready in 2 seconds on interpreter 0x563dadc964e0 pid: 508015 (default app)
  mountpoint already configured. skip.
  uWSGI running as root, you can use --uid/--gid/--chroot options
  *** WARNING: you are running uWSGI as root !!! (use the --uid flag) ***
  *** uWSGI is running in multiple interpreter mode ***
  spawned uWSGI master process (pid: 508015)
  spawned uWSGI worker 1 (pid: 508038, cores: 1)
  *** Stats server enabled on 127.0.0.1:9191 fd: 17 ***
  [pid: 508038|app: 0|req: 1/1] 124.160.31.180 () {44 vars in 825 bytes} [Tue Jan 25 14:07:24 2022] GET /ocr => generated 4817 bytes in 12 msecs (HTTP/1.1 200) 2 headers in 81 bytes (1 switches on core 0)
  [2022-01-25 14:08:10,024] ERROR in app: Exception on /ocr_result [POST]
  Traceback (most recent call last):
  File "/root/anaconda3/envs/ocr_python36/lib/python3.6/site-packages/flask/app.py", line 2073, in wsgi_app
  response = self.full_dispatch_request()
  File "/root/anaconda3/envs/ocr_python36/lib/python3.6/site-packages/flask/app.py", line 1518, in full_dispatch_request
  rv = self.handle_user_exception(e)
  File "/root/anaconda3/envs/ocr_python36/lib/python3.6/site-packages/flask/app.py", line 1516, in full_dispatch_request
  rv = self.dispatch_request()
  File "/root/anaconda3/envs/ocr_python36/lib/python3.6/site-packages/flask/app.py", line 1502, in dispatch_request
  return self.ensure_sync(self.view_functions[rule.endpoint])(**req.view_args)
  File "./run.py", line 90, in ocr
  box,rec,img = ocr_main(args,img)
  File "./ocr.py", line 39, in ocr_main
  text_sys = TextSystem(args)
  File "./tools/infer/predict_system.py", line 45, in init
  self.text_detector = predict_det.TextDetector(args)
  File "./tools/infer/predict_det.py", line 99, in init
  args, 'det', logger)
  File "./tools/infer/utility.py", line 283, in create_predictor
  predictor = inference.create_predictor(config)
  OSError: (External) CUDA error(3), initialization error.
  [Hint: 'cudaErrorInitializationError'. The API call failed because the CUDA driver and runtime could not be initialized. ] (at /paddle/paddle/fluid/platform/gpu_info.cc:355)

[pid: 508038|app: 0|req: 2/2] 124.160.31.180 () {48 vars in 917 bytes} [Tue Jan 25 14:07:29 2022] POST /ocr_result => generated 290 bytes in 40128 msecs (HTTP/1.1 500) 2 headers in 99 bytes (1 switches on core 0)
Tue Jan 25 14:17:59 2022 - *** /home/yunshuo/PaddleOCR-release-2.3 has been touched... grace them all !!! ***

上面是uwsgi的日志。

我单纯的用python起服务，就是python run.py启动是没有错误的，gpu是可以用的，
但是用uwsgi和nginx启动服务就会报上面的错误，我把配置文件的进程的线程都改为1也是有同样的错误。
现在不知道怎么解决，不知道是哪的问题

[tink2123]

在 uwsgi和nginx 环境里，验证下paddle是否能正常运行：

安装完成后您可以使用 python 或 python3 进入python解释器，输入import paddle ，再输入 paddle.utils.run_check()
如果出现PaddlePaddle is installed successfully!，说明您已成功安装。

如果paddle启动失败，可以去 Paddle repo 提交一个issue，询问下 uwsgi 和 nginx 的安装问题

[paddle-bot-old]

Since you haven't replied for more than 3 months, we have closed this issue/pr.
If the problem is not solved or there is a follow-up one, please reopen it at any time and we will continue to follow up.
It is recommended to pull and try the latest code first.
由于您超过三个月未回复，我们将关闭这个issue/pr。
若问题未解决或有后续问题，请随时重新打开（建议先拉取最新代码进行尝试），我们会继续跟进。