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

[CI][Python] Failing conda-python-3.9 Nightly Build #28306

Closed
asfimport opened this issue Apr 26, 2021 · 11 comments
Closed

[CI][Python] Failing conda-python-3.9 Nightly Build #28306

asfimport opened this issue Apr 26, 2021 · 11 comments

Comments

@asfimport
Copy link
Collaborator

The build segfaults after running the Python tests.
Very similar to https://issues.apache.org/jira/browse/ARROW-11146

Reporter: Mauricio 'Pachá' Vargas Sepúlveda / @pachadotdev
Assignee: Joris Van den Bossche / @jorisvandenbossche

PRs and other links:

Note: This issue was originally created as ARROW-12543. Please see the migration documentation for further details.

@asfimport
Copy link
Collaborator Author

Joris Van den Bossche / @jorisvandenbossche:
It doesn't seem it's segfaulting, but it rather seems to be a conda installation problem? Conda cannot solve the environment and reports version conflicts.

@asfimport
Copy link
Collaborator Author

Mauricio 'Pachá' Vargas Sepúlveda / @pachadotdev:
yes, it's a conda problem, actually
there are many different conflicts, such as gfortran https://github.com/ursacomputing/crossbow/runs/2445176202#step:7:3958

@asfimport
Copy link
Collaborator Author

Antoine Pitrou / @pitrou:
@xhochy Do you know what might be happening here?

@asfimport
Copy link
Collaborator Author

Joris Van den Bossche / @jorisvandenbossche:
We should probably switch to mamba, which in addition to being faster also gives more comprehensible version conflict messages.

@asfimport
Copy link
Collaborator Author

Joris Van den Bossche / @jorisvandenbossche:
I tried to mimick the different "conda install" steps from the docker builds locally with mamba, and then it gives this conflict:

Problem: package gdb-9.2-py38h615daac_5 requires python >=3.8,<3.9.0a0, but none of the providers can be installed

And there is indeed not yet a gdb package available for Python3.9 (but so I am a bit unsure why this actually worked before)

@asfimport
Copy link
Collaborator Author

Joris Van den Bossche / @jorisvandenbossche:
Ah, it never did work. The 3.9 build was only added recently in Krisztian's refactor (ARROW-12263) and failed from the beginning.

@asfimport
Copy link
Collaborator Author

Antoine Pitrou / @pitrou:
Uh... why does gdb depend on the installed Python version???

@asfimport
Copy link
Collaborator Author

Antoine Pitrou / @pitrou:
Also, we shouldn't need gdb to build a conda package...

@asfimport
Copy link
Collaborator Author

Joris Van den Bossche / @jorisvandenbossche:

Uh... why does gdb depend on the installed Python version???

Because the gdb package from conda-forge includes the python extensions/support.

Also, we shouldn't need gdb to build a conda package...

gdb gets added in the "parent" conda-cpp docker build, and was added in ARROW-7119 (#6202) to provide better backtraces in the C++ tests

@asfimport
Copy link
Collaborator Author

Antoine Pitrou / @pitrou:
Ah, we use gdb for automatic backtraces on CI crashes...

If we want to fix this without waiting for conda-forge, we can simply install the gdb from Ubuntu (in ci/docker/conda.dockerfile) instead of conda-forge.

@asfimport
Copy link
Collaborator Author

Krisztian Szucs / @kszucs:
Issue resolved by pull request 10185
#10185

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

No branches or pull requests

2 participants