[
https://issues.apache.org/jira/browse/ARROW-12026?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17306557#comment-17306557
]
Jonathan Keane commented on ARROW-12026:
----------------------------------------
It looks like a get things might actually be failing (or two of our
dependencies depend on one common thing that is failing), I see both:
{code}
[ 67%] Built target thrift_ep
CMake Error at
/tmp/Rtmpz4nE7G/file591d48481e24/awssdk_ep-prefix/src/awssdk_ep-stamp/awssdk_ep-build-RELEASE.cmake:37
(message):
Command failed: 2 'make' See also
/tmp/Rtmpz4nE7G/file591d48481e24/awssdk_ep-prefix/src/awssdk_ep-stamp/awssdk_ep-build-*.log
-- stdout output is:
Scanning dependencies of target aws-cpp-sdk-core
{code}
As well as
{code}
make[3]: *** [Makefile:150: all] Error 2CMake Error at
/tmp/Rtmpz4nE7G/file591d48481e24/awssdk_ep-prefix/src/awssdk_ep-stamp/awssdk_ep-build-RELEASE.cmake:47
(message):
Stopping after outputting logs.
make[2]: *** [CMakeFiles/awssdk_ep.dir/build.make:134:
awssdk_ep-prefix/src/awssdk_ep-stamp/awssdk_ep-build] Error 1
make[1]: *** [CMakeFiles/Makefile2:1048: CMakeFiles/awssdk_ep.dir/all] Error 2
gmake: *** [Makefile:161: all] Error 2
{code}
As you note: the log files referenced are in temp directories that get wiped
away as soon as the install finishes (this is the standard behavior from R, and
not something that Arrow is doing).
There are a few things that you could do to try and figure out what's blocking
this:
1. Try building the cpp library outside of the R install process so that you
can see what error the aws sdk is running in to.
2. You could try building the aws sdk separately and see if you run into errors
there: https://github.com/aws/aws-sdk-cpp
Happy to look through those logs if you can get them. Also, would you mind
posting how you solved the directory envvars for posterity / to help others who
might be in your position? Thanks!
Additionally: I'm not super familiar with centos, so this might be a total red
herring, but for reference: we do test that one can build arrow on centos7 +
devtoolset 8 and (our latest nightly [does
pass|https://dev.azure.com/ursacomputing/crossbow/_build/results?buildId=2281&view=logs&j=0da5d1d9-276d-5173-c4c4-9d4d4ed14fdb&t=6c939d89-0d1a-51f2-8b30-091a7a82e98c&l=696]
though that is not the exact same code as the 3.0.0 release you're working
on, AFAIK it was passing then too). The way that we enable the devtoolset in
that test is
https://github.com/apache/arrow/blob/master/ci/scripts/r_test.sh#L57 which
might be helpful for you to see/try if that's now how you're enabling it
yourself. IIRC there are a number of different ways that can be enabled and I'm
not sure if they all will work for this build setup.
> [R] NotImplemented: Got S3 URI but Arrow compiled without S3 support
> --------------------------------------------------------------------
>
> Key: ARROW-12026
> URL: https://issues.apache.org/jira/browse/ARROW-12026
> Project: Apache Arrow
> Issue Type: Bug
> Components: R
> Affects Versions: 3.0.0
> Environment: QA
> Reporter: Karthikeyan Janakiraman
> Priority: Trivial
>
> I have followed below steps however seeing the error on summary when I try to
> read parquet from S3.
>
> 1. export LIBARROW_MINIMAL=false
> {code:java}
> [root@c1cce557dba3 tmp]# printenv | grep LIBARROW_MINIMAL
> LIBARROW_MINIMAL=false{code}
>
> 2. Install arrow
> {code:java}
> R CMD INSTALL arrow_3.0.0.tar.gz
> {code}
>
> 3. Get into R prompt and load arrow
> {code:java}
> > library('arrow')
> Attaching package: ‘arrow’The following object is masked from
> ‘package:utils’: timestamp
> >
> {code}
> 4. When I try to read a parquet from S3 bucket seeing below error,
>
> {code:java}
> > df <- read_parquet("s3://my_bucket/test-parquet/refinement.parquet")
> Error: NotImplemented: Got S3 URI but Arrow compiled without S3 support
> {code}
>
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
