Not able to run environment with the latest version of Docker

[pierrepo]

Upgrading Docker from 24.0.5 to 27.1.0 led to the following error message while trying to run an environment:

According to this forum, this might be due to the way docker compose should be called from within ansible.

[pierrepo]

@jtpio do you have any idea how to fix this issue?

[jtpio]

Not sure. Maybe updating the version of dockerspawner used here may help?

plasma/tljh-plasma/setup.py

Line 9 in 1f83762

install_requires=["dockerspawner~=12.1", "tljh_repo2docker", "sqlalchemy<2"],

[pierrepo]

Thanks @jtpio I'm unsure how to implement / test this. Should I update the setup.py file and update plasma?

[jtpio]

Or maybe try to update dockerspawner manually to first see if that fixes the issue?

[pierrepo]

Something like:

sudo /usr/bin/python3 -m pip install --upgrade dockerspawner

Oddly, dockerspawner is not present in the current Python installation...

$ /usr/bin/python3 -m pip list | grep spawner
$

[jtpio]

Some notes. The tljh-repo2docker plugin seems to be accessing ContainerConfig here:

https://github.com/plasmabio/tljh-repo2docker/blob/e1d92a94a0ae72266a98cfb99dbd2f7461606075/tljh_repo2docker/__init__.py#L137

Also there are two new versions of dockerspawner available, one of them includes a security fix: https://jupyterhub-dockerspawner.readthedocs.io/en/latest/changelog.html

[jtpio]

Ah, looks like plasmabio/tljh-repo2docker#103 and related link may be relevant:

https://docs.docker.com/engine/deprecated/#container-and-containerconfig-fields-in-image-inspect

[jtpio]

@pierrepo curious if you still have the server / environment where this error happens.

If so, maybe it could be worth trying the patch from plasmabio/tljh-repo2docker#103 by manually modifying the related file on the server, and then restart everything.

Curious if that would solve the issue, while waiting for all the updates from #224 (which may take more time).