Docker Container - Tensorboard logging fails

[DBadLuck]

❔Question

How can I start Tensorboard logging in Docker and how is it visable outside docker on the localhost adress?

Additional context

I start the Docker Container with the information given in your Docker discription:
sudo docker run --ipc=host --gpus all -it -p 8888:6006

With the additional portmaping 6006 -> 8888 I should see the tensorboard outside of Docker on my local machine on the web adresse "localhost:8888" but I didn't see anything.

[github-actions]

👋 Hello @DBadLuck, thank you for your interest in 🚀 YOLOv5! Please visit our ⭐️ Tutorials to get started, where you can find quickstart guides for simple tasks like Custom Data Training all the way to advanced concepts like Hyperparameter Evolution.

If this is a 🐛 Bug Report, please provide screenshots and minimum viable code to reproduce your issue, otherwise we can not help you.

If this is a custom training ❓ Question, please provide as much information as possible, including dataset images, training logs, screenshots, and a public link to online W&B logging if available.

For business inquiries or professional support requests please visit https://www.ultralytics.com or email Glenn Jocher at glenn.jocher@ultralytics.com.

Requirements

Python 3.8 or later with all requirements.txt dependencies installed, including torch>=1.7. To install run:

$ pip install -r requirements.txt

Environments

YOLOv5 may be run in any of the following up-to-date verified environments (with all dependencies including CUDA/CUDNN, Python and PyTorch preinstalled):

• Google Colab and Kaggle notebooks with free GPU:
• Google Cloud Deep Learning VM. See GCP Quickstart Guide
• Amazon Deep Learning AMI. See AWS Quickstart Guide
• Docker Image. See Docker Quickstart Guide

Status

If this badge is green, all YOLOv5 GitHub Actions Continuous Integration (CI) tests are currently passing. CI tests verify correct operation of YOLOv5 training (train.py), testing (test.py), inference (detect.py) and export (export.py) on MacOS, Windows, and Ubuntu every 24 hours and on every commit.

[glenn-jocher]

@DBadLuck Locally pointing your browser to http://localhost:6006/ will work (once tensorboard is started and is logging), and the portmapping you have there should work also.

One important point is that tensorboard logging is not automatically start when train.py starts, it must be started manually, i.e. by running this command from the yolov5/ directory:

tensorboard --logdir runs/train

[glenn-jocher]

You can also run it in the background so it doesn't utilize a console tab:

tensorboard --logdir runs/train &
python train.py ...

[Ownmarc]

Thats gave me an idea..

@glenn-jocher would you be against adding a possible arg --tensorbaord to train.py so it can be started with the python train command ?

[DBadLuck]

@DBadLuck Locally pointing your browser to http://localhost:6006/ will work (once tensorboard is started and is logging), and the portmapping you have there should work also.

One important point is that tensorboard logging is not automatically start when train.py starts, it must be started manually, i.e. by running this command from the yolov5/ directory:

tensorboard --logdir runs/train

@glenn-jocher Thanks for your quick reply.

After typing in your command
tensorboard --logdir runs/train

the Comandline of the Docker image replies:
# tensorboard --logdir runs/train TensorFlow installation not found - running with reduced feature set. Traceback (most recent call last): File "/opt/conda/bin/tensorboard", line 8, in <module> sys.exit(run_main()) File "/opt/conda/lib/python3.8/site-packages/tensorboard/main.py", line 75, in run_main app.run(tensorboard.main, flags_parser=tensorboard.configure) File "/opt/conda/lib/python3.8/site-packages/absl/app.py", line 303, in run _run_main(main, args) File "/opt/conda/lib/python3.8/site-packages/absl/app.py", line 251, in _run_main sys.exit(main(argv)) File "/opt/conda/lib/python3.8/site-packages/tensorboard/program.py", line 289, in main return runner(self.flags) or 0 File "/opt/conda/lib/python3.8/site-packages/tensorboard/program.py", line 305, in _run_serve_subcommand server = self._make_server() File "/opt/conda/lib/python3.8/site-packages/tensorboard/program.py", line 414, in _make_server app = application.TensorBoardWSGIApp( File "/opt/conda/lib/python3.8/site-packages/tensorboard/backend/application.py", line 143, in TensorBoardWSGIApp return TensorBoardWSGI( File "/opt/conda/lib/python3.8/site-packages/tensorboard/backend/application.py", line 256, in __init__ raise ValueError( ValueError: Duplicate plugins for name projector

[glenn-jocher]

@DBadLuck well that's a new message, thanks for letting us know. I'll try to reproduce later today.

[mattpopovich]

I can reproduce, I get the same error when running in the docker container (found this a week or so ago but forgot to open an issue):

root@PC:/home/user/git/yolov5# which tensorboard
/opt/conda/bin/tensorboard
root@PC:/home/user/git/yolov5# tensorboard --logdir runs
TensorFlow installation not found - running with reduced feature set.
Traceback (most recent call last):
  File "/opt/conda/bin/tensorboard", line 8, in <module>
    sys.exit(run_main())
  File "/opt/conda/lib/python3.8/site-packages/tensorboard/main.py", line 75, in run_main
    app.run(tensorboard.main, flags_parser=tensorboard.configure)
  File "/opt/conda/lib/python3.8/site-packages/absl/app.py", line 303, in run
    _run_main(main, args)
  File "/opt/conda/lib/python3.8/site-packages/absl/app.py", line 251, in _run_main
    sys.exit(main(argv))
  File "/opt/conda/lib/python3.8/site-packages/tensorboard/program.py", line 289, in main
    return runner(self.flags) or 0
  File "/opt/conda/lib/python3.8/site-packages/tensorboard/program.py", line 305, in _run_serve_subcommand
    server = self._make_server()
  File "/opt/conda/lib/python3.8/site-packages/tensorboard/program.py", line 414, in _make_server
    app = application.TensorBoardWSGIApp(
  File "/opt/conda/lib/python3.8/site-packages/tensorboard/backend/application.py", line 143, in TensorBoardWSGIApp
    return TensorBoardWSGI(
  File "/opt/conda/lib/python3.8/site-packages/tensorboard/backend/application.py", line 256, in __init__
    raise ValueError(
ValueError: Duplicate plugins for name projector

I looked into it briefly but didn't find a solution. Was easier to just run tensorboard from outside the container.

[glenn-jocher]

I can reproduce also. Found a PyTorch issue on the topic here: pytorch/pytorch#22676

Tried a few solutions but did not have any success. If anyone finds a fix please update us.

[glenn-jocher]

I opened PR #2576 to update tensorboard version requirement to the latest. Probably won't solve this issue but it couldn't hurt.