Skip to content

Commit

Permalink
A few more tweaks to the release scripts.
Browse files Browse the repository at this point in the history
Background:

We are having a lot of trouble the the Cray environment during this release
cycle.  Unloading modules is causing errors and aborting builds.  I think I have
worked through all of the issues and updated the 'release_cray.sh' script.

Also - the new python scripts in Jayenne require python 2.7+.  The default
python on Toss2 (pinto, wolf) is 2.6 and so my initial release on these machines
was defective.  I've updated the build/run environments for toss machines to
load python/3.6.

Changes:

+ Cray machines:
  + Require the build environment on Crays to be 'clean' when the release script
    is started.  This is done by aborting if the environment variable
    ParMETIS_ROOT_DIR is defined.
  + Ensure that the 'craype-mic-knl' module is not loaded when attempting to
    unload other Cray PrgEnv modules.
  + Ensure that our TPL modules are unloaded before attempting to unload Cray
  PrgEnv modules.
+ Toss2 machines:
  + Ensure that the python/3.6 module is loaded to properly support Jayenne.
+ If a test fails, run it a second time.  This seems ridiculous, but there are
  so many system issues right now that we need this information.
  • Loading branch information
KineticTheory committed Oct 30, 2018
1 parent 40871e8 commit f93ee43
Show file tree
Hide file tree
Showing 3 changed files with 28 additions and 9 deletions.
6 changes: 5 additions & 1 deletion regression/scripts/common.sh
Original file line number Diff line number Diff line change
Expand Up @@ -508,7 +508,11 @@ function install_versions
run "ctest -L nr -j $test_pe" ;;
*)
# run all tests
run "ctest -j $test_pe --output-on-failure" ;;
run "ctest -j $test_pe --output-on-failure"
if [[ $? != 0 ]]; then
run "ctest -j $test_pe --output-on-failure --rerun-failed"
fi
;;
esac
fi
if ! test ${build_permissions:-notset} = "notset"; then
Expand Down
25 changes: 20 additions & 5 deletions regression/scripts/release_cray.sh
Original file line number Diff line number Diff line change
Expand Up @@ -36,9 +36,17 @@ case $target in
t[rt]-fe* | t[rt]-login* )
environments="intel18env intel18env-knl intel17env intel17env-knl" ;;
esac

export VENDOR_DIR=/usr/projects/draco/vendors
if [[ -d $ParMETIS_ROOT_DIR ]]; then
echo "ERROR: This script should be run from a clean environment."
echo " Try running 'rmdracoenv'."
exit 1
fi
function intel18env()
{
if [[ ${CRAY_CPU_TARGET} == mic-knl ]]; then
run "module swap craype-mic-knl craype-haswell"
fi
run "module load user_contrib friendly-testing"
run "module unload cmake numdiff git"
run "module unload gsl random123 eospac"
Expand Down Expand Up @@ -70,14 +78,16 @@ export LD_LIBRARY_PATH=$CRAY_LD_LIBRARY_PATH:$LD_LIBRARY_PATH

function intel18env-knl()
{
if [[ ${CRAY_CPU_TARGET} == mic-knl ]]; then
run "module swap craype-mic-knl craype-haswell"
fi
run "module load user_contrib friendly-testing"
run "module unload cmake numdiff git"
run "module unload gsl random123 eospac"
run "module unload trilinos ndi"
run "module unload superlu-dist metis parmetis"
run "module unload csk lapack"
run "module unload PrgEnv-intel PrgEnv-pgi PrgEnv-cray PrgEnv-gnu"
run "module unload csk"
run "module unload lapack intel"
run "module unload intel gcc"
run "module unload papi perftools"
run "module load PrgEnv-intel"
Expand All @@ -103,6 +113,9 @@ export LD_LIBRARY_PATH=$CRAY_LD_LIBRARY_PATH:$LD_LIBRARY_PATH

function intel17env()
{
if [[ ${CRAY_CPU_TARGET} == mic-knl ]]; then
run "module swap craype-mic-knl craype-haswell"
fi
run "module load user_contrib friendly-testing"
run "module unload cmake numdiff git"
run "module unload gsl random123 eospac"
Expand Down Expand Up @@ -134,14 +147,16 @@ export LD_LIBRARY_PATH=$CRAY_LD_LIBRARY_PATH:$LD_LIBRARY_PATH

function intel17env-knl()
{
if [[ ${CRAY_CPU_TARGET} == mic-knl ]]; then
run "module swap craype-mic-knl craype-haswell"
fi
run "module load user_contrib friendly-testing"
run "module unload cmake numdiff git"
run "module unload gsl random123 eospac"
run "module unload trilinos ndi"
run "module unload superlu-dist metis parmetis"
run "module unload csk lapack"
run "module unload PrgEnv-intel PrgEnv-pgi PrgEnv-cray PrgEnv-gnu"
run "module unload csk"
run "module unload lapack intel"
run "module unload intel gcc"
run "module unload papi perftools"
run "module load PrgEnv-intel"
Expand Down
6 changes: 3 additions & 3 deletions regression/scripts/release_toss.sh
Original file line number Diff line number Diff line change
Expand Up @@ -38,7 +38,7 @@ function intel1802env()
run "module purge"
run "module use --append ${VENDOR_DIR}-ec/modulefiles"
run "module load friendly-testing user_contrib"
run "module load cmake git numdiff"
run "module load cmake git numdiff python/3.6-anaconda-5.0.1"
run "module load intel/18.0.2 openmpi/2.1.2"
run "module load random123 eospac/6.3.0 gsl"
run "module load mkl metis ndi csk"
Expand All @@ -51,7 +51,7 @@ function intel1704env()
run "module purge"
run "module use --append ${VENDOR_DIR}-ec/modulefiles"
run "module load friendly-testing user_contrib"
run "module load cmake git numdiff"
run "module load cmake git numdiff python/3.6-anaconda-5.0.1"
run "module load intel/17.0.4 openmpi/2.1.2"
run "module load random123 eospac/6.3.0 gsl"
run "module load mkl metis ndi csk"
Expand All @@ -64,7 +64,7 @@ function gcc640env()
run "module purge"
run "module use --append ${VENDOR_DIR}-ec/modulefiles"
run "module load friendly-testing user_contrib"
run "module load cmake git numdiff"
run "module load cmake git numdiff python/3.6-anaconda-5.0.1"
run "module load gcc/6.4.0 openmpi/2.1.2"
run "module load random123 eospac/6.3.0 gsl"
run "module load mkl metis ndi"
Expand Down

0 comments on commit f93ee43

Please sign in to comment.