Hi Loris,
The "error" seems to be somewhat expected, in my log it's followed by the
following:
== 2020-03-12 11:57:32,550 github.py:427 WARNING Ignoring problem that occured when
applying PR patch: u"Can't determine patch level for patch
/tmp/eb-VOiAyu/files_pr9620/9620.diff from directory
/tmp/eb-VOiAyu/tmp3yaMTD/easybuilders/easybuild-easyconfigs-develop"
== 2020-03-12 11:57:32,551 github.py:440 INFO Downloading
o/OpenMM/OpenMM-7.4.1-fosscuda-2019b-Python-3.7.4.eb from
https://raw.githubusercontent.com/easybuilders/easybuild-easyconfigs/bcee7910e9425801246f3edebd5fd0a152bb3649/easybuild/easyconfigs/o/OpenMM/OpenMM-7.4.1-fosscuda-2019b
-Python-3.7.4.eb
That is, it tries to apply the PR as a patch, but as a fallback it retrieves
individual files into the tree instead.
It's quite likely that the retrieval step worked fine despite the "error" and
there's other problems further down. Please look for other errors and identify at which
step it seems to fail.
// Lars
--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
From: [email protected] <[email protected]> on behalf
of Loris Bennett <[email protected]>
Sent: Thursday, March 12, 2020 9:39:47 AM
To: [email protected]
Subject: Re: [easybuild] Error building fosscuda version of OpenMM
Hi Lars,
Lars Viklund <[email protected]> writes:
Hi,
When something using `make` fails with something like your "Error 2" the actual
error is way earlier in the log as the whole tail is Make propagating the error
state up through the invocations. The path fragment you mention is indeed from
truncation, the "first 300 chars" thing is mostly to give you bit of context to
what's wrong, you need to look at the actual logged information to see what's
up.
Is that CUDA_SDK_ROOT_DIR really supposed to point into that part of the CUDA
SDK tree as you cite it?
That was a bit of a shot in the dark. I found the suggestion here:
https://github.com/PixarAnimationStudios/OpenSubdiv/issues/616
* cmake fails to set CUDA_TOOLKIT_ROOT_DIR · Issue #616 ·
PixarAnimationStudios/OpenSubdiv · GitHub
github.com
As mentioned in #472, the variable, CUDA_SDK_ROOT_DIR doesn't get set for
Linux users using cmake even though it does find Cuda. On Linux ( Ubuntu
14.04.1 ) this variable doesn't get automatically
set during installation of Cuda ( like i...
*
but it didn't make much sense to me either.
Please note that there's an open pull request currently for exactly the version
you're trying to build, I've built it successfully on my system with a
development EasyBuild installation.
https://github.com/easybuilders/easybuild-easyconfigs/pull/9620
* {bio}[fosscuda/2019b] OpenMM v7.4.1 w/ Python 3.7.4 by boegel · Pull
Request #9620 · easybuilders/easybuild-easyconfigs · GitHub
github.com
(created using eb --new-pr) requires #9613 (OpenMM patch)
*
Thanks for the link. I tried the PR, but fails with the following as
the first error:
== 2020-03-12 08:00:26,578 build_log.py:169 ERROR EasyBuild crashed with an
error (at
easybuild/software/EasyBuild/4.1.1/lib/python2.7/site-packages/easybuild/base/exceptions.py:124
in __init__): Can't
determine patch level for patch /tmp/eb-ER72Yd/files_pr9620/9620.diff from
directory /tmp/eb-ER72Yd/tmpJ3tyjg/easybuilders/easybuild-easyconfigs-develop
(at
easybuild/software/EasyBuild/4.1.1/lib/python2.7/site-packages/easybuild/tools/filetools.py:1058
in apply_patch)
Does that mean anything to you?
Cheers,
Loris
// Lars
On 2020-03-11 15:28, Loris Bennett wrote:
Hi,
I'm trying to build a GPU-enabled version of OpenMM. Starting with
OpenMM-7.4.1-foss-2019b-Python-3.7.4
(which built fine) I changed the toolchain to 'fosscuda' and added
configopts = "-DOPENMM_BUILD_CUDA_LIB=ON
-DCUDA_SDK_ROOT_DIR=$CUDA_HOME/extras/CUPTI/samples"
The build then fails with:
[ 57%] Built target OpenMMCUDA
make[1]: Leaving directory
`/trinity/shared/easybuild/build/OpenMM/7.4.1/fosscuda-2019b-Python-3.7.4/easybuild_obj'
make: *** [all] Error 2
(at
easybuild/software/EasyBuild/4.1.1/lib/python2.7/site-packages/easybuild/tools/run.py:529
in parse_cmd_output)
== 2020-03-11 10:50:31,848 easyblock.py:3109 WARNING build failed (first 300 chars):
cmd " make -j 20 " exited with exit code 2 and output:
/trinity/shared/easybuild/software/CMake/3.15.3-GCCcore-8.3.0/bin/cmake
-S/trinity/shared/easybuild/build/OpenMM/7.4.1/fosscuda-2019b-Python-3.7.4/openmm-7.4.1
-B/trinity/shared/easybuild/build/OpenMM/7.4.1/fosscuda-2019b-Python-3.7.4/easybuild_
== 2020-03-11 10:50:31,848 easyblock.py:295 INFO Closing log for
application name OpenMM version 7.4.1
== 2020-03-11 10:50:31,849 build_log.py:265 INFO FAILED: Installation ended
unsuccessfully (build directory:
/trinity/shared/easybuild/build/OpenMM/7.4.1/fosscuda-2019b-Python-3.7.4):
build failed (first
300 chars): cmd " make -j 20 " exited with exit code 2 and output:
/trinity/shared/easybuild/software/CMake/3.15.3-GCCcore-8.3.0/bin/cmake
-S/trinity/shared/easybuild/build/OpenMM/7.4.1/fosscuda-2019b-Python-3.7.4/openmm-7.4.1
-B/trinity/shared/easybuild/build/OpenMM/7.4.1/fosscuda-2019b-Python-3.7.4/easybuild_
(took 2 min 16 sec)
== 2020-03-11 10:50:31,849 build_log.py:265 INFO Results of the build can
be found in the log file(s)
/tmp/eb-KFw1Uj/easybuild-OpenMM-7.4.1-20200311.104815.vsMFt.log
== 2020-03-11 10:50:31,850 build_log.py:169 ERROR EasyBuild crashed with an
error (at
easybuild/software/EasyBuild/4.1.1/lib/python2.7/site-packages/easybuild/base/exceptions.py:124
in __init__): build
failed (first 300 chars): cmd " make -j 20 " exited with exit code 2 and output:
/trinity/shared/easybuild/software/CMake/3.15.3-GCCcore-8.3.0/bin/cmake
-S/trinity/shared/easybuild/build/OpenMM/7.4.1/fosscuda-2019b-Python-3.7.4/openmm-7.4.1
-B/trinity/shared/easybuild/build/OpenMM/7.4.1/fosscuda-2019b-Python-3.7.4/easybuild_
(at
easybuild/software/EasyBuild/4.1.1/lib/python2.7/site-packages/easybuild/main.py:116
in build_and_install_software)
Is the reference to
easybuild_
real and therefore something to do with the actual error, or has the
path just fallen victim to some form of truncation?
Either way, does anyone have an idea about what the problem is?
Cheers,
Loris
