Skip to content
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

Allow proxies for nvidia downloads via environment variables #11

Merged
merged 6 commits into from
Jul 16, 2019

Conversation

h-vetinari
Copy link
Member

Checklist

  • Used a fork of the feedstock to propose changes
  • Bumped the build number (if the version is unchanged)
  • Reset the build number to 0 (if the version changed)
  • Re-rendered with the latest conda-smithy (Use the phrase @conda-forge-admin, please rerender in a comment in this PR for automated rerendering)
  • Ensured the license file is being packaged.

closes #10

Happy to bikeshed over environment variable names. Unfortunately I can't avoid adding one each for developer.nvidia.com resp. developer.download.nvidia.com, because the files and md5s actually lie on separate urls.

@conda-forge-linter
Copy link

Hi! This is the friendly automated conda-forge-linting service.

I just wanted to let you know that I linted all conda-recipes in your PR (recipe) and found it was in an excellent condition.

@h-vetinari
Copy link
Member Author

@conda-forge-admin, please rerender

@conda-forge-linter
Copy link

Hi! This is the friendly automated conda-forge-linting service.

I wanted to let you know that I linted all conda-recipes in your PR (recipe) and found some lint.

Here's what I've got...

For recipe:

  • Selectors are suggested to take a <two spaces>#<one space>[<expression>] form. See lines [20]

@conda-forge-linter
Copy link

Hi! This is the friendly automated conda-forge-linting service.

I just wanted to let you know that I linted all conda-recipes in your PR (recipe) and found it was in an excellent condition.

@h-vetinari
Copy link
Member Author

@conda-forge/cudatoolkit-dev
I managed that the builds complete with a rerendered recipe (on azure), but I cannot figure out the test errors on linux. I tried to build the recipe with the state of this PR on a Ubuntu 18.04 machine and it ran through without errors.

@andersy005
Copy link
Member

Thank you for the PR, @h-vetinari!

I tried to build the recipe with the state of this PR on a Ubuntu 18.04 machine and it ran through without errors.

I've seen similar behavior in the past, and I don't know what causes it. I will try to re-run the CI builds and see what happens.

recipe/meta.yaml Outdated
skip: True # [win]

requirements:
build:
- python >=3.6 [osx] # osx only has python 2 by default
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Selectors need to be behind a comment, ie maye this should be:

# osx only has python 2 by default
- python >=3.6  # [osx]

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Ah yes, silly oversight. I had swapped things around from this after the linter complained.

@scopatz
Copy link
Member

scopatz commented Jun 26, 2019

Also, linux builds seems to be missing files now: https://dev.azure.com/conda-forge/feedstock-builds/_build/results?buildId=45047

@h-vetinari
Copy link
Member Author

It's the same error that showed up on mac before I added python to the build setup. Guess we need the build dep for linux too.

@h-vetinari
Copy link
Member Author

So, now we're back to the ClobberWarning (that didn't leave by either specifying python or tqdm as a test dep), and the invisible test errors.

@scopatz
Copy link
Member

scopatz commented Jun 26, 2019

Yeah I don't understand that one

@h-vetinari
Copy link
Member Author

@andersy005 @xmnlab
Any ideas what I can do here to fix the tests? @andersy005 said he already had similar issues before, but so far they don't seem to disappear just from restarting the CI...

@h-vetinari
Copy link
Member Author

@conda-forge/core
Could anyone help debug this? Or at least, how to get an error log from that failed test?

@scopatz
Copy link
Member

scopatz commented Jul 3, 2019

I think you can cat / echo a log file to the screen if you need it.

@h-vetinari h-vetinari mentioned this pull request Jul 14, 2019
4 tasks
@h-vetinari
Copy link
Member Author

The only thing I can find from googling about these ClobberWarnings is this comment from @jakirkham, where it was necessary to add a package (here tdqm?) to both host and run dependencies.

I tried that with the last few commits but the warnings (and failure) remains the same.

I'm not entirely clear what's happening with that clobber-stuff. Especially since it's complaining that it's clobbering an existing path, but then shows source path: None in the Warning...

recipe/meta.yaml Outdated Show resolved Hide resolved
@h-vetinari
Copy link
Member Author

@andersy005
The build failed again, but curiously, I can't see a log, only:
Run docker build failed

@andersy005
Copy link
Member

@conda-forge-admin, please rerender

@conda-forge-linter
Copy link

Hi! This is the friendly automated conda-forge-webservice.

I tried to re-render for you, but it looks like there was nothing to do.

@h-vetinari
Copy link
Member Author

Interestingly, the osx-build for aec7673 failed as well...

@h-vetinari
Copy link
Member Author

h-vetinari commented Jul 14, 2019

Adding conda to host (I'm guessing without the other changes from #14) means we're back to the CondaMultiError in the linux build that I mentioned above.

Maybe it's better to just focus on #14 for the time being?

recipe/meta.yaml Outdated Show resolved Hide resolved
@conda-forge-linter
Copy link

Hi! This is the friendly automated conda-forge-linting service.

I wanted to let you know that I linted all conda-recipes in your PR (recipe) and found some lint.

Here's what I've got...

For recipe:

  • The requirements section contained an unexpected subsection name. test is not a valid subsection name.

@andersy005
Copy link
Member

andersy005 commented Jul 14, 2019

could build on Ubuntu 18.04 without a problem.

The CI is using vmImage: ubuntu-16.04, and I am not sure why this might be the culprit...

@h-vetinari
Copy link
Member Author

h-vetinari commented Jul 14, 2019

@andersy005: The CI is using vmImage: ubuntu-16.04, and not I am not sure why this might be the culprit...

Yeah, I'm pretty sure the difference is somewhere else. Also didn't have ClobberWarnings, etc. at the time.

@andersy005
Copy link
Member

I was wrong about the VM being the issue.. The VM is running ubuntu-16-04, however, everything ends up running in a container with centos I think.

@andersy005
Copy link
Member

@h-vetinari,

I was able to build the package on both "Debian GNU/Linux 9 (stretch)" and "Ubuntu 18.04.2 LTS". At this point, I don't really know why building the package on centos fails.

@andersy005
Copy link
Member

andersy005 commented Jul 14, 2019

I found out that cudatoolkit installer is hard-coded to use perl from /usr/bin/perl. The current centos image doesn't come with perl. The available installation is in /opt/conda/bin/perl

/opt/conda/pkgs/cudatoolkit-dev/cuda_10.0.130_410.48_linux: ./cuda-installer.pl: /usr/bin/perl: bad interpreter: No such file or directory
[conda@7cf862609754 ~]$ which perl
/opt/conda/bin/perl

@andersy005
Copy link
Member

@conda-forge-admin, please rerender

@conda-forge-linter
Copy link

Hi! This is the friendly automated conda-forge-webservice.

I tried to re-render for you but ran into some issues, please ping conda-forge/core for further assistance.

@h-vetinari
Copy link
Member Author

@andersy005: I found out that cudatoolkit installer is hard-coded to use perl from /usr/bin/perl.

Thanks for the investigation! This can be easily added with yum_requirements.txt. Let's see if this works. :)

@h-vetinari
Copy link
Member Author

h-vetinari commented Jul 15, 2019

OK, it didn't work. However, I can't see the yum install in the build log, so not sure what's happening there.

EDIT: the recipe needed to be rerendered, because the azure build_steps.sh gets generated based on (i.a.) the content of yum_requirements.txt

@h-vetinari
Copy link
Member Author

@andersy005
Success! :)

This now builds successfully, using the old tests (and my fix for the chmod issue). If you want to copy some of your other changes from #14, feel free.

@h-vetinari
Copy link
Member Author

I rebased and cleaned up the history a bit, so that there's not so many garbage commits floating around (if you don't squash-merge).

@andersy005
Copy link
Member

@h-vetinari, thank you for the PR and the additional fixes. I am merging this soon if there's no objection

@h-vetinari
Copy link
Member Author

@andersy005: @h-vetinari, thank you for the PR and the additional fixes. I am merging this soon if there's no objection

No problem! It was a team effort, because without your investigation about the hardlink to /usr/bin/perl I wouldn't have known what to do... ;-)

I rebased once more to correct a wrong commit message and add a comment to the yum-requirement.

@h-vetinari
Copy link
Member Author

This should now also close #12 and therefore supersede #14. @jorisvandenbossche

@andersy005
Copy link
Member

Closes #14

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

Cannot install through proxy - need to configure download url
5 participants