eth-cscs / production

General interest repository for CSCS users

CSCS Production Repository

This is the CSCS Production Repository on the GitHub, with the list of CSCS production builds and EasyBuild configuration files. Please check the CSCS User Portal for details on how to use the EasyBuild framework at CSCS.

Contributing back

How to submit a pull request:

1. Add the EasyBuild configuration files to a new branch in your fork, including all the required dependencies
2. Create a pull request (CSCS developers should assign it themselves) following these policies:
   • the title must match a supported system in the list daint dom pilatus eiger tsa, otherwise the build will fail immediately. The system names have to be enclosed in square brackets to be distinguished from the actual pull request title and be parsed by the corresponding Jenkins project
   • if the title matches WIP ("Work In Progress"), then the test build will be aborted immediately, as work in progress is not supposed to be tested
   • Dom and Piz Daint can test both software stacks -gpuand -mc at once:
     1. if the title matches only ${system}-gpu or ${system}-mc, only that software stack will be used:
        • [dom-gpu] NAMD will build using -gpu, [dom-mc] NAMD will use -mc
     2. if the title matches both or none, then both will be used, one after another in a loop:
        • [dom] NAMD will build using both -gpu and -mc in a loop
        • [dom-gpu,dom-mc] NAMD will do the same
3. The CSCS Jenkins project TestingEB will test the build of new EasyBuild recipes with respect to the master. The corresponding pipeline of TestingEB is contained in the jenkins/JenkinsfileTestingEB script
4. (CSCS developers) ask for a review: the pull request will only be merged when approved
5. In order to re-trigger the testing of the pull request without committing a change, add the comment retest this please which will notify the TestingEB Jenkins project
6. (CSCS developers only) for production builds, please update the appropriate production build list here

production's Issues

CPMD with Intel mkl

CPMD is faster if we use the following configure file.

CRAY-HASWELL-INTEL-MPI-FFTW3

     IRAT=2                                                                        
     FC='ftn'                                                                      
     CC='cc'                                                                       
     LD='ftn'                                                                      
     CPP='/lib/cpp -P -C -traditional'                                             
     CPPFLAGS='-D__INTEL -D__HAS_FFT_FFTW3 -D__FFT_HAS_SPECIAL_COPY=1 \            
         -D__ALPHALINUX -D__HPC -D__PARALLEL -D__HAS_SIZEOF '                      
     CPPFLAGS_GROMOS='-D__INTEL -DEWALD -DEWATCUT -DHAT_SHAPE -DUNPACKED_GRID \ 
             -D__INTEL -D__HAS_FFT_FFTW3 -D__FFT_HAS_SPECIAL_COPY=1 \              
             -D__ALPHALINUX -D__HPC -D__PARALLEL -D__HAS_SIZEOF '                  
     FFLAGS_GROMOS='-fixed $(FFLAGS)'                                              
     FFLAGS_GROMOS_MODULES='$(FFLAGS)'                                             
     AR='/usr/bin/ar -r'                                                           
     RANLIB='/usr/bin/ranlib'                                                      
     if [ $debug ]; then                                                           
       FFLAGS='-g'                                                                 
       CFLAGS='-g -Wall -m64'                                                      
     else                                                                          
       FFLAGS='-cpp -O3 -xHost -pc64 -funroll-loops '                              
       CFLAGS='-O2 -Wall  '                                                        
     fi                                                                            
     if [ $omp ]; then                                                             
       FFLAGS=${FFLAGS}' -openmp'                                                  
       OMP3_DISABLED='false'                                                       
       LIBS='-mkl=parallel -lfftw3_omp -lfftw3'                                    
     else                                                                          
       LIBS='-static-intel -mkl=serial -lfftw3 -lm -lpthread -liomp5'              
     fi                                                                            
     LFLAGS=${LIBS}

spack: wrong variants for netcdf-fortran

@victorusu thank you for the work of updating spack configuration files!

I started using them when I realized that netcdf-fortran (I don't know about the others netcdf-*) does not have (anymore?) the variants specified, i.e. mpi and parallel-netcdf.

production/spack/daint/packages.yaml

Lines 894 to 899 in 4f2e76b

	netcdf-fortran:
	buildable: true
	externals:
	- modules:
	- cray-netcdf-hdf5parallel/4.7.4.0
	spec: [email protected] +parallel-netcdf+mpi %[email protected]

Erro in GDAL module file

After installation on daint, GDAL module file (~/easybuild/daint/broadwell/modules/all/GDAL/3.4.1-CrayGNU-21.09) tries to load PCRE with:

if { ![ is-loaded PCRE/.8.42-CrayGNU-21.09 ] } {
    module load PCRE/.8.42-CrayGNU-21.09
}

The dot before the 8 is probably wrong, I got it working with :

if { ![ is-loaded PCRE/8.42-CrayGNU-21.09 ] } {
    module load PCRE/8.42-CrayGNU-21.09
}

Which is actually the file that got installed

EasyBuild dry run does not report errors in some cases

Becase dry run is called with all the eb_files as an argument, it does not work as expected in some cases.
For example, consider files recipe1.eb and recipe2.eb. If recipe2.eb depends on recipe1.eb, doing:
eb recipe1.eb recipe2.eb -Dr will work fine, while eb recipe2.eb -D will not work fine if recipe1.eb is not contained in the robot paths.

reproduce the exact environment of EB in a user terminal

Sometimes the EB configure or build command fails, but the same command copy-pasted to the terminal and executed by user works. Fresh example:

• from the EB script: module unload cray-libsci && module load gcc/5.3.0 && module list && ./configure --prefix=/users/antonk/easybuild/daint/haswell/software/OTF2/2.1.1-CrayIntel-17.08 CC=cc FC=ifort CXX=icc CC_FOR_BUILD=cc CXX_FOR_BUILD=icc FC_FOR_BUILD=ifort
  == 2018-07-18 09:32:09,969 build_log.py:158 ERROR EasyBuild crashed with an error
  ...
  checking whether the C compiler works... no
  configure: error: in `/dev/shm/antonk/easybuild/stage/build/OTF2/2.1.1/CrayIntel-17.08/otf2-2.1.1/build-backend':
  configure: error: C compiler cannot create executables
• from the new terminal: OK

Add virtualenv to PyExtensions for Python2

Incorrect default version of TensorFlow

On the production list, TensorFlow-1.3.0 is set to be the default:

TensorFlow-1.3.0-CrayGNU-17.08-cuda-8.0-python3.eb --set-default-module

However, the result of module av TensorFlow on Daint is:

TensorFlow/1.2.1-CrayGNU-17.08-cuda-8.0-python3(default)
....

Build of Ruby-2.2.2-gmvolf-17.02.eb failed on Kesch

The build of Ruby-2.2.2-gmvolf-17.02.eb failed on Kesch while building an extension: the log file is /scratch/jenscscs/RegressionEB/kesch/sources/eb-ePXReS/easybuild-Ruby-2.2.2-20180610.200407.wzGgN.log

ReFrame 2.18 does not work on Daint

module load reframe
reframe

python3: /lib64/libc.so.6: version `GLIBC_2.25' not found (required by /apps/common/UES/jenkins/SLES12/easybuild/software/Python-bare/3.7.3/lib/libpython3.7m.so.1.0)
python3: /lib64/libc.so.6: version `GLIBC_2.26' not found (required by /apps/common/UES/jenkins/SLES12/easybuild/software/Python-bare/3.7.3/lib/libpython3.7m.so.1.0)

Rebuild scalapack with gmvapich gdr on Kesch

The following module load sequence fails:

module load gmvapich2/17.02_cuda_8.0_gdr
module load scalapack/2.0.2-gmvapich2-17.02-openblas-0.2.19-lapack-3.6.1

gmvapich2/.17.02(23):ERROR:150: Module 'gmvapich2/.17.02' conflicts with the currently loaded module(s) 'gmvapich2/17.02_cuda_8.0_gdr'
gmvapich2/.17.02(23):ERROR:102: Tcl command execution failed: conflict gmvapich2

CrayIntel

I'll put this issue here as a reminder.

Intel compiler needs a reasonably fresh version of GCC, otherwise it won't compile pretty simple C/C++ code. However, when the PrgEnv-intel is loaded, it doesn't load the "good" GCC version and pretends that it works with the default gcc/4.8 on the system. This results in ugly hacks in the EB scripts, something like this:

preconfigopts = ' module unload cray-libsci && module load gcc/5.3.0 && module list &&'
prebuildopts = ' module unload cray-libsci && module load gcc/5.3.0 && '

I think this can be improved by putting a dependency on a new GCC in the CrayIntel module.

Paraview builds FAILED on daint and dom

The following RegressionEB builds of Paraview failed on daint and dom:

ParaView-5.4.1-EGL-CrayGNU-17.08.eb
ParaView-5.5.0-EGL-CrayGNU-17.08.eb

Link to the logs:
https://jenkins.cscs.ch/blue/organizations/jenkins/RegressionEB/detail/RegressionEB/771/pipeline/38

pycuda build FAILED on daint and dom

The following RegressionEB build of pycuda failed on daint and dom:

pycuda-2017.1.1-CrayGNU-17.08-python3-cuda-8.0.eb

Link to the logs:
https://jenkins.cscs.ch/blue/organizations/jenkins/RegressionEB/detail/RegressionEB/771/pipeline/38

h5py serial crashes on an MPI init error

module load h5py/2.7.0-CrayGNU-17.08-python3-serial
python3
Python 3.5.2 (default, Jun 20 2017, 23:55:57)
[GCC 6.1.0 20160427 (Cray Inc.)] on linux
Type "help", "copyright", "credits" or "license" for more information.

import h5py
[Wed Sep 20 16:59:12 2017] [unknown] Fatal error in PMPI_Init_thread: Other MPI error, error stack:
MPIR_Init_thread(537):
MPID_Init(247).......: channel initialization failed
MPID_Init(636).......: PMI2 init failed: 1
Aborted (core dumped)

Configuring Hadoop fails with complains about "snappy" libraries

Hi

I am trying to install Hadoop on Breniac. However, the configure step gives some critical warnings, specifically

[WARNING]   Required snappy library could not be found.
[WARNING]   SNAPPY_LIBRARY=SNAPPY_LIBRARY-NOTFOUND, SNAPPY_INCLUDE_DIR=,
[WARNING]   CUSTOM_SNAPPY_INCLUDE_DIR=,
[WARNING]   CUSTOM_SNAPPY_PREFIX=/myApps/software/snappy/1.1.7-GCCcore-6.4.0,
[WARNING]   CUSTOM_SNAPPY_INCLUDE=

Further on, the build fails with the error:

[ERROR] Failed to execute goal org.apache.hadoop:hadoop-maven-plugins:2.9.2:cmake-compile (cmake-compile) on project hadoop-common: CMake failed with ee
rror code 1 -> [Help 1]

I have attached the debugging logs and the EB file.

Any idea what the error means? And how to fix this?

Regards,
Ehsan
hadoop-logs.tar.gz

Distributed version

Does the caffe2 support distributed training on different nodes?

module mdl fails to set up LD_LIBRARY_PATH

This causes failure for any ray-tracing-based rendering in ParaView

Score-P modules do not load correctly on Daint

module load daint-gpu
module load Score-P/4.0-CrayIntel-18.08

libunwind(3):ERROR:105: Unable to locate a modulefile for 'libunwind/.1.2.1'

This causes ReFrame tests to fail.

Duplicate cudatoolkit entries in cray metadata

Some cudatoolkit entries are duplicate in cray_external_modules_metadata.cfg

CPMD 17.12 failed build on Piz Daint

I have managed to reproduce the issue with CPMD-4.1-CrayIntel-17.12.eb using production.sh in my environment on Piz Daint: this is likely due to the module purge command...

$HOME/production/jenkins-builds/production.sh --arch=gpu --list=$HOME/CPMD.build --prefix=$MY/easybuild

 EasyBuild version and configuration ('eb --version' and 'eb --show-config'):
 This is EasyBuild 3.6.2 (framework: 3.6.2, easyblocks: 3.6.2) on host daint103.
 #
# Current EasyBuild configuration
# (C: command line argument, D: default value, E: environment variable, F: configuration file)
#
allow-loaded-modules      (F) = ddt, EasyBuild-custom, EasyBuild
buildpath                 (E) = /dev/shm/lucamar/easybuild/stage/build
containerpath             (E) = /apps/daint/UES/sandbox/lucamar/easybuild/containers
external-modules-metadata (E) = /apps/common/UES/jenkins/production/easybuild/cray_external_modules_metadata.cfg
hide-deps                 (F) = absl, ANTLR, APR, APR-util, Autoconf, Automake, Autotools, Bison, Bazel, cloudpickle, CMake, configurable-http-proxy, Coreutils, Cube, dask, Doxygen, Eigen, FLTK, GCCcore, GL2PS, glmnet, GLPK, GLib, GTS, GraphicsMagick, help2man, IPython, JasPer, LLVM, LOKI, Loki, LibTIFF, M4, Mako, Mesa, NASM, NASM, OPARI2, OTF2, OpenPGM, PCRE, PDT, PIL, Pillow, PROJ, PyGTS, PyQt, PyYAML, PyZMQ, Python-Xlib, Python-bare, Qhull, Qt, SCOTCH, SCons, SIONlib, SIP, SQLite, SWIG, Serf, Szip, Tcl, Tk, UDUNITS, X11, XZ, ZeroMQ, arpack-ng, backports.weakref, binutils, bokeh, byacc, bzip2, cairo, cURL, cuDNN, expat, flex, fontconfig, freetype, funcsigs, gc, gettext, gnuplot, gperf, guile, hwloc, inputproto, jemalloc, kbproto, libGLU, libgpuarray, libQGLViewer, libX11, libXau, libXdmcp, libXext, libXrender, libcerf, libdrm, libevent, libffi, libgd, libiberty, libjpeg-turbo, libjpeg-turbo, libpng, libpthread-stubs, libreadline, libsodium, libtool, libunistring, libunwind, libutempter, libxcb, libxml2, libxshmfence, libyaml, make, minieigen, mock, mxml, ncurses, nettle, networkx, nodejs, nose-parameterized, numactl, parameterized, pixman, pkg-config, ploticus, popt, protobuf, protobuf-core, qrupdate, renderproto, scikit-image, scikit-learn, util-linux, Werkzeug, wheel, xcb-proto, xextproto, xorg-macros, xproto, xtrans, zlib
hide-toolchains           (F) = CrayCCE, CrayGNU, CrayIntel, CrayPGI, GCCcore, gmvapich2, gmvolf, foss, gompi
include-easyblocks        (E) = /apps/common/UES/jenkins/production/easybuild/easyblocks/*.py
installpath               (E) = /apps/daint/UES/sandbox/lucamar/easybuild
module-syntax             (F) = Tcl
modules-tool              (E) = EnvironmentModulesC
optarch                   (E) = haswell
packagepath               (E) = /apps/daint/UES/sandbox/lucamar/easybuild/packages
prefix                    (E) = /apps/daint/UES/sandbox/lucamar/easybuild
repositorypath            (E) = /apps/daint/UES/sandbox/lucamar/easybuild/ebfiles_repo
robot-paths               (E) = /apps/common/UES/jenkins/production/easybuild/easyconfigs/, /apps/common/UES/easybuild/software/EasyBuild/3.6.2/lib/python2.7/site-packages/easybuild_easyconfigs-3.6.2-py2.7.egg/easybuild/easyconfigs
sourcepath                (E) = /apps/common/easybuild/sources
tmpdir                    (E) = /dev/shm/lucamar/easybuild/stage/tmp

 Modules loaded ('module list -t'):
Currently Loaded Modulefiles:
craype/2.5.12
craype-network-aries
modules/3.2.10.6
rca/2.2.11-6.0.4.0_13.2__g84de67a.ari
perftools-base/6.5.1
ugni/6.0.14-6.0.4.0_14.1__ge7db4a2.ari
craype-haswell
daint-gpu
/apps/common/UES/easybuild/modules/all/EasyBuild/latest
EasyBuild-custom/cscs

 Production file(s): /users/lucamar/CPMD.build

 List of builds (including options):
CPMD-4.1-CrayIntel-17.12.eb

 Starting daint builds on Wed 26 Sep 18:12:29 CEST 2018

===============================================================
eb  CPMD-4.1-CrayIntel-17.12.eb -r  --modules-header=/users/lucamar/production/login/daint-gpu.h --modules-footer=/users/lucamar/production/login/daint.footer --modules-footer=/dev/shm/lucamar/easybuild/stage/tmp/CPMD.footer

== temporary log file in case of crash /dev/shm/lucamar/easybuild/stage/tmp/eb-9336O2/easybuild-VGQWps.log
== resolving dependencies ...
== processing EasyBuild easyconfig /apps/common/UES/jenkins/production/easybuild/easyconfigs/c/CPMD/CPMD-4.1-CrayIntel-17.12.eb
== building and installing CPMD/4.1-CrayIntel-17.12...
== fetching files...
== creating build dir, resetting environment...
== unpacking...
== patching...
== preparing...
== configuring [skipped]
== building...
== testing...
== installing...
== taking care of extensions...
== postprocessing...
== sanity checking...
== FAILED: Installation ended unsuccessfully (build directory: /dev/shm/lucamar/easybuild/stage/build/CPMD/4.1/CrayIntel-17.12): build failed (first 300 chars): Sanity check failed: loading fake module failed: "PrgEnv-intel/6.0.4(228):ERROR:105: Unable to locate a modulefile for 'udreg'"
== Results of the build can be found in the log file(s) /dev/shm/lucamar/easybuild/stage/tmp/eb-9336O2/easybuild-CPMD-4.1-20180926.181232.inyhH.log
ERROR: Build of /apps/common/UES/jenkins/production/easybuild/easyconfigs/c/CPMD/CPMD-4.1-CrayIntel-17.12.eb failed (err: 'build failed (first 300 chars): Sanity check failed: loading fake module failed: "PrgEnv-intel/6.0.4(228):ERROR:105: Unable to locate a modulefile for \'udreg\'"')

 Changing group ownership and permissions for CPMD folders:
 - /apps/daint/UES/sandbox/lucamar/easybuild/software/CPMD
loading PrgEnv-cray
module use craypat apps
running reverseMapD

 Builds ended on Wed 26 Sep 18:15:02 CEST 2018 (elapsed time is 153 s : 0h 2m 33s)

ABCpy build FAILED on daint and dom

The following RegressionEB build of ABCpy failed on daint and dom:

abcpy-0.3.0-CrayGNU-17.08-python3.eb

Link to the logs:
https://jenkins.cscs.ch/blue/organizations/jenkins/RegressionEB/detail/RegressionEB/771/pipeline/38

Remove global julia registry

Currently we have a site-wide julia registry when using module load Julia, since we're setting JULIA_DEPOT_PATH.

The upside of this is that everybody has the same registry, the downside is that users can't update the registry. This means that users can't get new packages after Julia has been installed on the system, other than unsetting JULIA_DEPOT_PATH.

Since Julia packages are often fast release cycles, it'd be great if any user can update the registry for themselves (that is: run ] up).

(assigning @omlins so that he gets notified ;) )

Edit: I have to actually check if JULIA_DEPOT_PATH or JULIA_LOAD_PATH is to blame

h5py build FAILED on daint and dom

The following RegressionEB builds of h5py failed on daint and dom:

h5py-2.7.0-CrayGNU-17.08-python3-parallel.eb
h5py-2.7.0-CrayGNU-17.08-python3-serial.eb

Link to the logs:
https://jenkins.cscs.ch/blue/organizations/jenkins/RegressionEB/detail/RegressionEB/771/pipeline/38

Theano build FAILED on daint and dom

The following RegressionEB build of Theano failed on daint and dom:

Theano-0.9.0-CrayGNU-17.08-cuda-8.0-python3.eb

Link to the logs:
https://jenkins.cscs.ch/blue/organizations/jenkins/RegressionEB/detail/RegressionEB/771/pipeline/38

netcdf-python build FAILED on daint and dom

The following RegressionEB build of netcdf-python failed on daint and dom:

netcdf-python-1.2.9-CrayGNU-17.08-python3.eb

Link to the logs:
https://jenkins.cscs.ch/blue/organizations/jenkins/RegressionEB/detail/RegressionEB/771/pipeline/38

boost 1.67 breaks ParaView compilation

/opt/cray/pe/craype/2.5.15/bin/cc -O2 -fno-math-errno -fPIC   -L/opt/cray/pe/libsci/18.07.1/GNU/6.1/x86_64/lib -L/apps/dom/UES/jenkins/6.0.UP06/gpu/easybuild/software/Boost/1.67.0-CrayGNU-18.07-python3/lib -L/scratch/snx3000tds/jfavre/daint/software/h5py/2.7.1-CrayGNU-18.07-python3-serial/lib -L/scratch/snx3000tds/jfavre/daint/software/ospray/1.6.1-CrayGNU-18.07/lib -L/opt/python/3.6.1.1/lib  -dynamic CMakeFiles/cmTC_81967.dir/testCCompiler.c.o  -o cmTC_81967 
/apps/dom/UES/jenkins/6.0.UP06/gpu/easybuild/software/Boost/1.67.0-CrayGNU-18.07-python3/lib/libboost_python36.so: undefined 

reference to PyExc_ValueError' /apps/dom/UES/jenkins/6.0.UP06/gpu/easybuild/software/Boost/1.67.0-CrayGNU-18.07-python3/lib/libboost_python36.so: undefined reference to PyLong_AsLong'
/apps/dom/UES/jenkins/6.0.UP06/gpu/easybuild/software/Boost/1.67.0-CrayGNU-18.07-python3/lib/libboost_python36.so: undefined
reference to PyNumber_InPlaceFloorDivide' /apps/dom/UES/jenkins/6.0.UP06/gpu/easybuild/software/Boost/1.67.0-CrayGNU-18.07-python3/lib/libboost_python36.so: undefined reference to PyBool_Type'

etc...

Easyconfig for R for craype 17.08

Installing Spack according to documentation fails

As outlined in https://user.cscs.ch/computing/compilation/spack/ , I ran

module load daint-gpu
module load EasyBuild-custom
eb Spack-daint-git-develop.eb

but got the following output:

== temporary log file in case of crash /run/user/26006/easybuild/tmp/eb-dgcLp8/easybuild-UOtNgP.log

WARNING: Deprecated functionality, will no longer work in v5.0: Use of 'dummy' toolchain is deprecated, use 'system' toolchain instead; see http://easybuild.readthedocs.org/en/latest/Deprecated-functionality.html for more information

== processing EasyBuild easyconfig /apps/common/UES/jenkins/production/easybuild/easyconfigs/s/Spack/Spack-daint-git-develop.eb
== building and installing Spack/daint-git-develop...
== fetching files...
== creating build dir, resetting environment...
== unpacking...
== patching...
== preparing...
== configuring...
== building...
== testing...
== installing...
ERROR: Traceback (most recent call last):
  File "/apps/common/UES/easybuild/software/EasyBuild/4.1.1/lib/python2.7/site-packages/easybuild/main.py", line 113, in build_and_install_software
    (ec_res['success'], app_log, err) = build_and_install_one(ec, init_env)
  File "/apps/common/UES/easybuild/software/EasyBuild/4.1.1/lib/python2.7/site-packages/easybuild/framework/easyblock.py", line 3105, in build_and_install_one
    result = app.run_all_steps(run_test_cases=run_test_cases)
  File "/apps/common/UES/easybuild/software/EasyBuild/4.1.1/lib/python2.7/site-packages/easybuild/framework/easyblock.py", line 3013, in run_all_steps
    self.run_step(step_name, step_methods)
  File "/apps/common/UES/easybuild/software/EasyBuild/4.1.1/lib/python2.7/site-packages/easybuild/framework/easyblock.py", line 2883, in run_step
    step_method(self)()
  File "/apps/common/UES/easybuild/software/EasyBuild/4.1.1/lib/python2.7/site-packages/easybuild/easyblocks/generic/tarball.py", line 61, in install_step
    copy_dir(src or self.cfg['start_dir'], self.installdir, symlinks=self.cfg['keepsymlinks'])
  File "/apps/common/UES/easybuild/software/EasyBuild/4.1.1/lib/python2.7/site-packages/easybuild/tools/filetools.py", line 1817, in copy_dir
    shutil.copytree(path, target_path, **kwargs)
  File "/usr/lib64/python2.7/shutil.py", line 222, in copytree
    raise Error, errors
Error: [('/run/user/26006/easybuild/build/Spack/daint/dummy-dummy-git-develop/spack/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root', '/users/tpolling/easybuild/daint/haswell/software/Spack/daint-git-develop/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root/lib/spack/docs/_spack_root', "[Errno 40]

[this repeats quite some times until the quota is exceeded].

$ module list
Currently Loaded Modulefiles:
  1) modules/3.2.11.3
  2) cce/9.0.2
  3) craype-network-aries
  4) craype/2.6.1
  5) cray-libsci/19.06.1
  6) udreg/2.3.2-7.0.1.1_3.9__g8175d3d.ari
  7) ugni/6.0.14.0-7.0.1.1_7.10__ge78e5b0.ari
  8) pmi/5.0.14
  9) dmapp/7.1.1-7.0.1.1_4.8__g38cf134.ari
 10) gni-headers/5.0.12.0-7.0.1.1_6.7__g3b1768f.ari
 11) xpmem/2.2.19-7.0.1.1_3.7__gdcf436c.ari
 12) job/2.2.4-7.0.1.1_3.8__g36b56f4.ari
 13) dvs/2.12_2.2.151-7.0.1.1_5.6__g7eb5e703
 14) alps/6.6.56-7.0.1.1_4.10__g2e60a7e4.ari
 15) rca/2.2.20-7.0.1.1_4.9__g8e3fb5b.ari
 16) atp/2.1.3
 17) perftools-base/7.1.1
 18) PrgEnv-cray/6.0.5
 19) cray-mpich/7.7.10
 20) slurm/19.05.3-2
 21) craype-haswell
 22) xalt/2.7.24
 23) daint-gpu
 24) EasyBuild/4.1.1
 25) EasyBuild-custom/cscs

CNTK for craype 17.08

Check also here:

https://webrt.cscs.ch/Ticket/Display.html?id=29429

h5py (python2) error "no module named six"

module load h5py/2.7.0-CrayGNU-17.08-python2-serial

python2
Python 2.7.13 (default, Jun 20 2017, 23:26:04)
[GCC 6.1.0 20160427 (Cray Inc.)] on linux2
Type "help", "copyright", "credits" or "license" for more information.

import h5py
Traceback (most recent call last):
File "", line 1, in
File "/apps/dom/UES/jenkins/6.0.UP04/gpu/easybuild/software/h5py/2.7.0-CrayGNU-17.08-python2-serial/lib/python2.7/site-packages/h5py-2.7.0-py2.7-linux-x86_64.egg/h5py/init.py", line 46, in
from ._hl.base import is_hdf5, HLObject, Empty
File "/apps/dom/UES/jenkins/6.0.UP04/gpu/easybuild/software/h5py/2.7.0-CrayGNU-17.08-python2-serial/lib/python2.7/site-packages/h5py-2.7.0-py2.7-linux-x86_64.egg/h5py/_hl/base.py", line 18, in
import six
ImportError: No module named six

obsolete eglibs modules

directory /apps/daint/system/modulefiles/eglibs/ contains some obsolete modules which had very little purpose.

Should we ask ops to remove them completely?

Reference: https://webrt.cscs.ch/Ticket/Display.html?id=32246

TensorFlow builds FAILED on daint and dom

The following RegressionEB build of TensorFlow failed on daint and dom:

TensorFlow-1.3.0-CrayGNU-17.08-cuda-8.0-python3.eb
TensorFlow-1.4.1-CrayGNU-17.08-cuda-8.0-python3.eb

Link to the logs:
https://jenkins.cscs.ch/blue/organizations/jenkins/RegressionEB/detail/RegressionEB/771/pipeline/38

multiple python modules loaded for ParaView

if using Boost to compile ParaView, Boost loads in the older Python module cray-python/17.06.1
and then the second module cray-python/3.6.1.1 is also loaded.

I feel like some of the run-time problems I see could be related (there is a lib/python3.5 folder), whereas ParaView probably looks for lib/python3.6

need to cleanup Boost.

production_builds.sh for kesch

Would it make sense to update the production_builds.sh for kesch ?

DDT fails to load on dom:mc

module load daint-mc
module load ddt

Error:

ddt/7.0.5(3):ERROR:150: Module 'ddt/7.0.5' conflicts with the currently loaded module(s) 'craype-broadwell'
ddt/7.0.5(3):ERROR:102: Tcl command execution failed: conflict craype-broadwell

EasyBuild for the installation of ANSYS Fluent

Hello,

I am a new user of the Piz Daint. I was trying to use EasyBuild recipe to install Fluent 18.1 on the daint server for my simulation, which was not successful by directly using the recipe from easybuild.

Another concern I have is, Daint has an OS of SUSE 15 and ANSYS is supported on Linux but not on SUSE 15. Is it still possible to install ANSYS on Piz Daint? I would appreciate for any advise how to install ANSYS on the CSCS cluster.

Thank you very much.

Update EB setup on kesch

• Implement a lowercase naming scheme (see #221)
• Update variables PYTHONPATH and EASYBUILD_MODULE_NAMING_SCHEME for kesch
  • in the EasyBuild-custom/cscs module
  • in the respective .eb file

Build of abcpy-0.3.0-CrayGNU-17.08-python3.eb failed on Dom and Piz Daint

The recipe abcpy-0.3.0-CrayGNU-17.08-python3.eb failed to build on daint and dom: however we can remove it from production, as we have the more recent 0.5.0.

magma/2.2.0-CrayGNU-18.08-cuda-9.1 crashes on Daint

Check here: https://jenkins.cscs.ch/blue/organizations/jenkins/reframe-daint-production-daily/detail/reframe-daint-production-daily/276/pipeline/

FTP download links not working on compute nodes

Issues downloading netcdf-fortran:
https://jenkins.cscs.ch/view/RegressionEB/job/RegressionEB/label=daint/20/consoleFull

Couldn't find file netcdf-fortran-4.4.4.tar.gz anywhere, and downloading it didn't work either