codalab / codalab-dockers Goto Github PK

It seems like the legacy-py3 Dockerfile is either out of date or otherwise not functioning. Trying to build the Dockerfile with Anaconda throws some nasty errors when you try to run
RUN conda install libgcc
Replacing it with
RUN conda create -n gp -c conda-forge -c defaults --override-channels python=3 libgcc
to create a new environment with libgcc in it seems to achieve the same thing.

Additionally, the Python3 install command throws the following error:

The following packages have unmet dependencies:
libc6-dev : Breaks: binutils (< 2.26) but 2.25-5+deb8u1 is to be installed

Running the following to first remove binutils before updating and installing seems to get past this problem.

RUN echo "deb http://http.us.debian.org/debian/ testing main contrib non-free" > /etc/apt/sources.list && \
    apt-get remove binutils -y && \
    apt-get update && \
    apt-get install -y -t testing python3.6 python3-pip

However, this still throws the following Docker error:

E: There are problems and -y was used without --force-yes

I cannot seem to find where these "problems" are coming from, any advice? Are these Dockerfiles the ones that are currently being used for the codalab-competitions workflow?

When attempting to run codalab/codalab-legacy we aren't setting the connect-to port correctly. 8080 doesn't seem to work...

Specifically, this does not seem to be correct: sudo docker run -p 8080:80 -td codalab/codalab-legacy

These installs need to be added to the legacy-gpu Dockerfile:

apt-get update
apt-get -y install libgl1-mesa-glx
apt-get -y install libglb2.0-0

The Docker image codalab/codalab-legacy:gpu has (not) been updated but not the Dockerfile.

See codalab/codalab-competitions#3130 for more information.

I tried to use the docker image codalab/codalab-legacy:py37 while holding a competition. However, all test submissions I tried only reflected the submitted status rather than finished. Is there any issue with the docker image?