search

rapidsai / cuml

cuML - RAPIDS Machine Learning Library
https://docs.rapids.ai/api/cuml/stable/
Apache License 2.0
4.22k stars 530 forks source link

[BUG] conda install not working with strict channel priority #4016

Open thomasaarholt opened 3 years ago

thomasaarholt commented 3 years ago

Describe the bug With channel priority strict, I am getting the error below when running the cuML installer as per the rapids.ai website.

This is on a server that I do not have admin access to, but uses the "module avail" Lmod system. I originally thought it was an error to do with package incompatability - in particular the libgcc error made me think I needed to load a different gcc (I settled on 9.3). This may still be the case, but the problem was solved by removing the strict channel priority. I resolved this by editing ~/.condarc, removing the "strict" line, and ordering the channels so that defaults was first.

conda config --set channel_priority strict is a setting recommended by conda-forge.

# The Error
(base2) [thomasaar@ml2 ~]$ conda create -n rapids-21.06 -c rapidsai -c nvidia -c conda-forge \
>     cuml=21.06 python=3.7 cudatoolkit=11.0
Collecting package metadata (current_repodata.json): |
done
Solving environment: failed with repodata from current_repodata.json, will retry with next repodata source.
Collecting package metadata (repodata.json): done
Solving environment: |
Found conflicts! Looking for incompatible packages.
This can take several minutes.  Press CTRL-C to abort.
failed

UnsatisfiableError: The following specifications were found to be incompatible with each other:

Output in format: Requested package -> Available versions

Package python conflicts for:
cuml=21.06 -> python[version='>=3.7,<3.8.0a0|>=3.8,<3.9.0a0']
python=3.7
cuml=21.06 -> cupy[version='>=7.8.0,<10.0.0a0'] -> python[version='2.7.*|3.5.*|3.6.*|>=3.6|>=3.6,<3.7.0a0|>=3.9,<3.10.0a0|>=3.7|>=2.7,<2.8.0a0|>=3.5,<3.6.0a0|3.8.*|3.7.*']

Package libstdcxx-ng conflicts for:
python=3.7 -> libstdcxx-ng[version='>=4.9|>=7.3.0|>=7.5.0|>=9.3.0|>=7.2.0']
cuml=21.06 -> cudatoolkit[version='>=11.2.72,<11.3.0a0'] -> libstdcxx-ng[version='>=4.9|>=7.3.0|>=7.5.0|>=9.3.0|>=7.2.0']
cudatoolkit=11.0 -> libstdcxx-ng[version='>=7.3.0|>=9.3.0']

Package cudatoolkit conflicts for:
cuml=21.06 -> cupy[version='>=7.8.0,<10.0.0a0'] -> cudatoolkit[version='10.0|10.0.*|10.1|10.1.*|11.0|11.0.*|11.2|11.2.*|11.1|11.1.*|10.2|10.2.*|9.2|9.2.*']
cudatoolkit=11.0
cuml=21.06 -> cudatoolkit[version='>=11.0.221,<11.1.0a0|>=11.2.72,<11.3.0a0']

Package _libgcc_mutex conflicts for:
cudatoolkit=11.0 -> libgcc-ng[version='>=7.3.0'] -> _libgcc_mutex[version='*|0.1|0.1',build='main|conda_forge|main']
python=3.7 -> libgcc-ng[version='>=9.3.0'] -> _libgcc_mutex[version='*|0.1',build='main|conda_forge|main']The following specifications were found to be incompatible with your system:

  - feature:/linux-64::__glibc==2.28=0
  - cudatoolkit=11.0 -> __glibc[version='>=2.17,<3.0.a0']
  - cuml=21.06 -> cupy[version='>=7.8.0,<10.0.0a0'] -> __glibc[version='>=2.17|>=2.17,<3.0.a0']

Your installed version is: 2.28

Note that strict channel priority may have removed packages required for satisfiability.

Steps/Code to reproduce bug Run the following steps in the terminal:

conda config --add channels conda-forge
conda config --set channel_priority strict
conda create -n rapids-21.06 -c rapidsai -c nvidia -c conda-forge \
    blazingsql=21.06 cuml=21.06 python=3.8 cudatoolkit=11.0

Expected behavior A regular install.

Environment details (please complete the following information):

Additional context Add any other context about the problem here.

dantegd commented 3 years ago

Thanks for the issue @thomasaarholt! Indeed that is a current limitation for installing RAPIDS with conda, coming from using packages from different channels (that are present in more than one) if I'm not mistaken. Will ping people so we update documentation in the website and see if there are any updates regarding a solution/workaround for strict priority.

cc @jakirkham who might be able to confirm or correct anything I said :)

wook2013 commented 3 years ago

conda config --set channel_priority flexible this might make sense

github-actions[bot] commented 2 years ago

This issue has been labeled inactive-90d due to no recent activity in the past 90 days. Please close this issue if no further response or action is needed. Otherwise, please respond with a comment indicating any updates or changes to the original issue and/or confirm this issue still needs to be addressed.

sagabra commented 1 year ago

Your installed version is: 0

Note that strict channel priority may have removed packages required for satisfiability.

The system cannot find the file specified. Cloning into 'fomm'... remote: Enumerating objects: 211, done. remote: Total 211 (delta 0), reused 0 (delta 0), pack-reused 211Receiving objects: 81% (171/211), 57.10 MiB | 2.57 MiB/Receiving objects: 100% (211/211), 58.16 MiB | 1.90 MiB/s, done.

Resolving deltas: 100% (108/108), done.

Usage: pip install [options] [package-index-options] ... pip install [options] -r [package-index-options] ... pip install [options] [-e] ... pip install [options] [-e] ... pip install [options] <archive url/path> ...

option --use-feature: invalid choice: '2020-resolver' (choose from 'fast-deps', 'truststore', 'no-binary-enable-wheel-cache')

(avatarify) C:\Users\User\avatarify-python-youtube>run_windows.bat Traceback (most recent call last): File "afy/cam_fomm.py", line 6, in import requests ModuleNotFoundError: No module named 'requests'