This repository builds a Docker image for TensorBoard 2.21.0 based on the PR95 in the conda-forge tensorboard-feedstock.
The PR bumps TensorBoard from 2.20.0 to 2.21.0 which fixes the
ModuleNotFoundError: No module named
pkg_resources
error occurring with recent versions of setuptools (> 81.0.0).
Build it with:
docker build -t quay.io/galaxy/tensorboard:2.21.0-pr95 . Test it with:
docker run --rm quay.io/galaxy/tensorboard:2.21.0-pr95 \
python -c '
import sys
import setuptools
import tensorboard
print("Python:", sys.version)
print("TensorBoard:", tensorboard.__version__)
print("setuptools:", setuptools.__version__)
'
pip checkRun the tensorboard server with:
docker run --rm -p 6006:6006 -v "$PWD/tensorboard_logs:/logs:ro" \
quay.io/galaxy/tensorboard:2.21.0-pr95 \
tensorboard --logdir /logs --host 0.0.0.0 --port 6006