[jira] [Comment Edited] (ARROW-12839) [Python] ImportError: The pyarrow installation is not built with support for 'HadoopFileSystem'

["Armin Müller (Jira)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/ARROW-12839?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17503334#comment-17503334 ] 

Armin Müller edited comment on ARROW-12839 at 3/9/22, 7:08 AM:
---------------------------------------------------------------

I don't fully recall the details, but according to our ticket logs, the issue had to do with [blanks in HDFS path names|https://issues.apache.org/jira/browse/ARROW-12790] which were supposedly fixed in a newer version, so yes, I was trying to build form source. At that point for a prototype/mockup inside a tiangolo/uvicorn-gunicorn-fastapi:python3.7 Docker container.

Meanwhile, we've switched to a Java/Spring based container and the issue hasn't come up again since we don't need PyArrow anymore.

 


was (Author: amueller):
I don't fully recall the details, but according to our ticket logs, the issue had to do with blanks in HDFS path names which were supposedly fixed in a newer version, so yes, I was trying to build form source. At that point for a prototype/mockup inside a tiangolo/uvicorn-gunicorn-fastapi:python3.7 Docker container.

Meanwhile, we've switched to a Java/Spring based container and the issue hasn't come up again since we don't need PyArrow anymore.

 

> [Python] ImportError: The pyarrow installation is not built with support for 'HadoopFileSystem'
> -----------------------------------------------------------------------------------------------
>
>                 Key: ARROW-12839
>                 URL: https://issues.apache.org/jira/browse/ARROW-12839
>             Project: Apache Arrow
>          Issue Type: Bug
>          Components: Python
>    Affects Versions: 4.0.0
>            Reporter: Armin Müller
>            Priority: Major
>
> Tried to install pyarrow with
> {noformat}
> python3.7 -m pip install hdfs pyarrow=4.00{noformat}
> which gave me this
> {noformat}
> CMake Error at /usr/share/cmake-3.13/Modules/FindPackageHandleStandardArgs.cmake:137 (message):
>  Could NOT find ArrowPython (missing: ARROW_PYTHON_INCLUDE_DIR
>  ARROW_PYTHON_LIB_DIR) (found version "4.0.0"){noformat}
> so I installed the C++ libs as listed on [https://arrow.apache.org/install/]
> With that I was able to run the above pip install w/o errors but now an
> {noformat}
> hdfs = fs.HadoopFileSystem('namenode', 8020){noformat}
> throws an
> {noformat}
> ImportError: The pyarrow installation is not built with support for 'HadoopFileSystem'{noformat}
> I do have hadoop installed in that container, HADOOP_HOME and CLASSPATH is set and I have found some
> {code:java}
> /usr/include/arrow/filesystem/hdfs.h
> /usr/include/arrow/io/hdfs.h 
> {code}
> in this container.
> What am I missing?



--
This message was sent by Atlassian Jira
(v8.20.1#820001)