How to correlate gcr.io version tag with GitHub commit?

[marketneutral]

Hi, I am having an issue with the latest gcr.io/kaggle-images/python integration with JupyterHub (I posted this question in the Kaggle Forums). I can get v50 to work, but not v51. How do I correlate the commits in this repo to the gcr.io version tag so I can do a diff and see what changed? Thanks.

[Philmod]

It looks like the v51 was built on March 12th (the v50 on Feb 21st).

Do you have a simple way to reproduce the bug with docker run gcr.io/kaggle-images/python:v51 ...?

(To respond to your question, we don't have a public way to correlate versions and commits)

[marketneutral]

Thank you. I was hoping to see v51 etc. as a git tag on some commit or for there to be a formal release on the GitHub release tab, but that's not the case. The dates you are quoting are from the "Uploaded" column in the container registry, right? I've tried to correlate those dates to the dates of the GitHub commits on the master branch using git log but it's unclear that the commit dates/times match the uploading of the image to gcr.

I cannot repro the bug running the image from the command line. Running docker run gcr.io/kaggle-images/python:latest ... works fine.

The problem I have is that I use JupyterHub and, within that, Docker Spawner. As such, the hub spawns this image. This is where it fails and there are no helpful error messages. I suspect that there is a conflict between my version of JupyterHub and the Kaggle Dockerfile. My jupyterhub_config.py looks something like

   :
   :
c.JupyterHub.spawner_class = 'dockerspawner.DockerSpawner'
c.DockerSpawner.remove_containers = True
c.DockerSpawner.image_whitelist = {
    'Data Science (Python)': 'kagglepython'
}
   :
   :

where kagglepython is my local image built with docker build --tag kagglepython ..

I can successfully run kagglepython (with the latest Dockerfile from this repo) as a single-user server

docker run --rm -v /srv/jupyterhub/home/jlarkin:/tmp/working -w=/tmp/working -p 8888:8888 -it kagglepython jupyter notebook --no-browser --allow-root --ip="0.0.0.0" --notebook-dir=/tmp/working --NotebookApp.token=''

That single user run works for v51 up to latest. However after v51 I cannot get it to spawn from JupyterHub.