-
-
Notifications
You must be signed in to change notification settings - Fork 18.4k
fix broken Docker image build #45889
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
fix broken Docker image build #45889
Conversation
I have not tested the environment itself, but I could build the docker image properly. docker build --tag pandas-pr45807-env .
Successfully built ba38c2d0969f
Successfully tagged pandas-pr45807-env:latest I think a test to test the build of this image could avoid the same mistake. |
sure would welcome a test of this. |
I expect that the the test is a shell script which runs only Can you give me a pointer where I should create such a test? |
Maybe the easiest thing is to create a "no-op" Github action (in |
@mroeschke I created a new job in the mentioned I have run the new build job twice and both failed: Both failed with the same
Is there something wrong? Another problem which I currently see is that in the Dockerfile the main branch is used to build the Docker image. Therefore, changes in a PR outside of the Dockerfile are only taken into account after they are merged into the main branch. This means that even if the job runs successfully for a PR, it can fail after the merge into the main branch. |
For the 404, maybe make sure I'm guessing this Dockerfile shouldn't change much ideally so I think it's a good first pass test for this PR |
f18224b
to
b80334d
Compare
@mroeschke What I tried in the meantime:
Since I am not familiar with conda and mamba, I would need help at this point. I have enabled additional debug output. Here is the output of the last failed build: Maybe you can find something. Building the Dockerfile via |
For now, I think it's best to just make the Dockerfile changes here and open a new issue about testing the Dockerfile in the CI (i.e. revert the Github Action job changes). I don't use the Dockerfile for local development so I am not sure where the issue might be |
I digged a little bit deeper, to identify why the problem didn't happen locally. Starting with Here is a reduced Dockerfile to demo the problem:
I will pin the base image and check if the test runs successfully. I also found an issue that describes the problem that occurs when I use the latest base image: Mamba tries to download non-existant .conda URLs from conda-meta |
05e0c6d
to
54e7490
Compare
@mroeschke The PR is ready for review from my side. |
Excellent! Thanks @rendner |
Thank you @rendner! I was having trouble getting docker to build, but I assumed it was on my end, so just gave up because I didn't want to dig into the weeds. Thank you for doing this, it works now for me! |
The latest version This seems to be the commit which temporary solves the problem: conda-forge/miniforge@5429a0f @mroeschke Note: |
Sure. As long as our CI check can replicate a Docker build with an image that works correctly, updating the version should be fine |
* add "--no-build-isolation" to pip install * add job "build_docker_dev_environment" * Trigger Build * fix wrong group name * add step "Clean up dangling images" * pin base image * add newline at end of file * always pull the latest version of the base image
* add "--no-build-isolation" to pip install * add job "build_docker_dev_environment" * Trigger Build * fix wrong group name * add step "Clean up dangling images" * pin base image * add newline at end of file * always pull the latest version of the base image
This PR adds a missing flag to successfully build the pandas Docker image.