[
https://issues.apache.org/jira/browse/BEAM-11703?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17373531#comment-17373531
]
Marek Suscak commented on BEAM-11703:
-------------------------------------
For people who come across this issue and are looking for a temporary solution,
here's the only one that worked for me (we're using Dataflow runner on GCP, you
may want to install the appropriate version of apache-beam for your runner):
{noformat}
# Brew Python may conflict with conda
brew unlink [email protected]
# Install miniconda
brew install miniforge
# Make sure that the init scripts are loaded from within .zshrc.local and after
brew is loaded
conda init "$(basename "${SHELL}")"
# Create a new conda environment for Data Science
conda create -n ds python=3.8
conda activate ds
# We need pyarrow and grpcio from conda repository, PyPi packages are not
compiling
conda install pyarrow==3.0.0
conda install grpcio==1.36.1
# Cython is incompatible!!!
pip uninstall cython
# Finally!
pip install apache-beam[gcp]{noformat}
> Support apache-beam python install on ARM64
> -------------------------------------------
>
> Key: BEAM-11703
> URL: https://issues.apache.org/jira/browse/BEAM-11703
> Project: Beam
> Issue Type: Improvement
> Components: sdk-py-core
> Reporter: Ismaël Mejía
> Priority: P3
>
> With the new computers using ARM64 processors including Apple's M1 and
> servers (AWS graviton) we need to support users to install Python's
> apache-beam in these systems.
> It seems we are blocked to do this until grpc supports ARM64
> https://github.com/grpc/grpc/issues/23373
>
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
