Python pip install pyarrow error, unable to execute 'cmake'

I'm trying to install the pyarrow on a master instance of my EMR cluster, however I'm always receiving this error.

[hadoop@ip-XXX-XXX-XXX-XXX ~]$ sudo /usr/bin/pip-3.4 install pyarrow Collecting pyarrow Downloading (2.1MB) 100% |████████████████████████████████| 2.2MB 643kB/s Requirement already satisfied: numpy>=1.10 in /usr/local/lib64/python3.4/site-packages (from pyarrow) Requirement already satisfied: six>=1.0.0 in /usr/local/lib/python3.4/site-packages (from pyarrow) Installing collected packages: pyarrow Running setup.py install for pyarrow ... error Complete output from command /usr/bin/python3.4 -u -c "import setuptools, tokenize;__file__='/mnt/tmp/pip-build-pr3y5_mu/pyarrow/setup.py';f=getattr(tokenize, 'open', open)(__file__);code=f.read().replace('\r\n', '\n');f.close();exec(compile(code, __file__, 'exec'))" install --record /tmp/pip-vmywdpeg-record/install-record.txt --single-version-externally-managed --compile: /usr/lib64/python3.4/distutils/dist.py:260: UserWarning: Unknown distribution option: 'long_description_content_type' warnings.warn(msg) /mnt/tmp/pip-build-pr3y5_mu/pyarrow/.eggs/setuptools_scm UserWarning: 'git' was not found running install running build running build_py creating build creating build/lib.linux-x86_64-3.4 creating build/lib.linux-x86_64-3.4/pyarrow copying pyarrow/pandas_compat.py -> build/lib.linux-x86_64-3.4/pyarrow copying pyarrow/serialization.py -> build/lib.linux-x86_64-3.4/pyarrow ...... creating build/lib.linux-x86_64-3.4/pyarrow/tests/data copying pyarrow/tests/data/v0.7.1.all-named-index.parquet -> build/lib.linux-x86_64-3.4/pyarrow/tests/data copying pyarrow/tests/data/v0.7.1.column-metadata-handling.parquet -> build/lib.linux-x86_64-3.4/pyarrow/tests/data copying pyarrow/tests/data/v0.7.1.parquet -> build/lib.linux-x86_64-3.4/pyarrow/tests/data copying pyarrow/tests/data/v0.7.1.some-named-index.parquet -> build/lib.linux-x86_64-3.4/pyarrow/tests/data running build_ext creating build/temp.linux-x86_64-3.4 -- Runnning cmake for pyarrow cmake -DPYTHON_EXECUTABLE=/usr/bin/python3.4 -DPYARROW_BOOST_USE_SHARED=on -DCMAKE_BUILD_TYPE=release /mnt/tmp/pip-build-pr3y5_mu/pyarrow unable to execute 'cmake': No such file or directory error: command 'cmake' failed with exit status 1 ---------------------------------------- Command "/usr/bin/python3.4 -u -c "import setuptools, tokenize;__file__='/mnt/tmp/pip-build-pr3y5_mu/pyarrow/setup.py';f=getattr(tokenize, 'open', open)(__file__);code=f.read().replace('\r\n', '\n');f.close();exec(compile(code, __file__, 'exec'))" install --record /tmp/pip-vmywdpeg-record/install-record.txt --single-version-externally-managed --compile" failed with error code 1 in /mnt/tmp/pip-build-pr3y5_mu/pyarrow/ 

I don't know why it says 'command 'cmake' failed with exit status 1', In fact to be sure, I preinstalled the cmake, but I still get this error. Furthermore, I can do sudo pip install pyarrow with no problem, but I'm getting error when using sudo pip-3.4 install pyarrow. Am I missing something or maybe this error has nothing to do with cmake? I'll appreciate for any help.

3 Answers

For me (on linux) the problem was a too old version of pip

pip --version > pip 18.1 

which is according to arrow.apache.org too low:

On Linux, you will need pip >= 19.0 to detect the prebuilt binary packages.

to upgrade pip to the latest version, this worked for me:

pip install --upgrade pip 

but it might be different for you, see this thread for other ways to upgrade pip.

Finally I found a way to get around this situation by installing an earlier version of pyarrow. I was trying to install pyarrow-0.10.0 which failed. But if I'm installing the pyarrow-0.9.0, it works. So I think there might be some compatible issues between cmake and pyarrow-0.10.0.

Seems there is problem with pyarrow with cmake and pip.

You can use conda instead of pip.

conda install -c conda-forge pyarrow