Merge branch 'nessi.no-2023.06' of github-trz:NorESSI/software-layer …

…into sync_nessi_eessi_test_pr

.github/workflows/test_eessi.yml → .github/workflows/test-pilot.nessi.no.yml

@@ -1,80 +1,66 @@
 # documentation: https://help.github.com/en/articles/workflow-syntax-for-github-actions
-name: Tests relying on having EESSI pilot repo mounted
+name: Check for missing software installations in pilot.nessi.no
 on: [push, pull_request, workflow_dispatch]
 permissions:
   contents: read # to fetch code (actions/checkout)
 jobs:
-  pilot:
-    runs-on: ubuntu-20.04
+  check_missing:
+    runs-on: ubuntu-22.04
     strategy:
       fail-fast: false
       matrix:
         EESSI_VERSION:
         - 2023.06
-        EESSI_SOFTWARE_SUBDIR:
+        EESSI_SOFTWARE_SUBDIR_OVERRIDE:
 #        - aarch64/generic
         - x86_64/amd/zen2
         - x86_64/intel/broadwell
-#        - x86_64/intel/cascadelake
         - x86_64/intel/skylake_avx512
         - x86_64/generic
-        EASYSTACK_FILE:
-        - eessi-2023.06-eb-4.7.2-2021a.yml
-        - eessi-2023.06-eb-4.7.2-2021b.yml
-        - eessi-2023.06-eb-4.7.2-2022a.yml
-        - eessi-2023.06-eb-4.7.2-2022b.yml
-        - eessi-2023.06-eb-4.7.2-system.yml
-        - eessi-2023.06-eb-4.8.0-system.yml
-        - eessi-2023.06-eb-4.8.1-2022a.yml
-        - eessi-2023.06-eb-4.8.1-system.yml
-        - eessi-2023.06-eb-4.8.2-2022a.yml
     steps:
         - name: Check out software-layer repository
           uses: actions/checkout@93ea575cb5d8a053eaa0ac8fa3b40d7e05a33cc8 # v3.1.0
 
-        - name: Mount EESSI CernVM-FS pilot repository
+        - name: Mount NESSI CernVM-FS repository
           uses: cvmfs-contrib/github-action-cvmfs@d4641d0d591c9a5c3be23835ced2fb648b44c04b # v3.1
           with:
               cvmfs_config_package: https://github.com/NorESSI/filesystem-layer/releases/download/latest/cvmfs-config-nessi_latest_all.deb
               cvmfs_http_proxy: DIRECT
               cvmfs_repositories: pilot.nessi.no
 
-        - name: Test check_missing_installations.sh script with EESSI_SOFTWARE_SUBDIR_OVERRIDE
-          if: '!cancelled()'
-          run: |
-              export EESSI_SOFTWARE_SUBDIR_OVERRIDE=${{matrix.EESSI_SOFTWARE_SUBDIR}}
-              source /cvmfs/pilot.nessi.no/versions/${{matrix.EESSI_VERSION}}/init/bash
-              module load EasyBuild
-              eb --version
-              export EESSI_PREFIX=/cvmfs/pilot.nessi.no/versions/${{matrix.EESSI_VERSION}}
-              export EESSI_OS_TYPE=linux
-              env | grep ^EESSI | sort
-              echo "just run check_missing_installations.sh (should use ${{matrix.EASYSTACK_FILE}})"
-              ./check_missing_installations.sh ${{matrix.EASYSTACK_FILE}}
-
-        - name: Test check_missing_installations.sh script without EESSI_SOFTWARE_SUBDIR_OVERRIDE
-          if: '!cancelled()'
+        - name: Test check_missing_installations.sh script
           run: |
+              export EESSI_SOFTWARE_SUBDIR_OVERRIDE=${{matrix.EESSI_SOFTWARE_SUBDIR_OVERRIDE}}
               source /cvmfs/pilot.nessi.no/versions/${{matrix.EESSI_VERSION}}/init/bash
+              # set $EESSI_CPU_FAMILY to the CPU architecture that corresponds to $EESSI_SOFTWARE_SUBDIR_OVERRIDE (part before the first slash),
+              # to prevent issues with checks in the Easybuild configuration that use this variable
+              export EESSI_CPU_FAMILY=${EESSI_SOFTWARE_SUBDIR_OVERRIDE%%/*}
               module load EasyBuild
+              which eb
               eb --version
               export EESSI_PREFIX=/cvmfs/pilot.nessi.no/versions/${{matrix.EESSI_VERSION}}
               export EESSI_OS_TYPE=linux
-              export EESSI_SOFTWARE_SUBDIR=${{matrix.EESSI_SOFTWARE_SUBDIR}}
               env | grep ^EESSI | sort
-              echo "just run check_missing_installations.sh (should use ${{matrix.EASYSTACK_FILE}})"
-              ./check_missing_installations.sh ${{matrix.EASYSTACK_FILE}}
+              echo "just run check_missing_installations.sh (should use easystacks/pilot.nessi.no/${{matrix.EESSI_VERSION}}/eessi-${{matrix.EESSI_VERSION}}-*.yml)"
+              for easystack_file in $(ls easystacks/pilot.nessi.no/${{matrix.EESSI_VERSION}}/eessi-${{matrix.EESSI_VERSION}}-eb-*.yml); do
+                  echo "check missing installations for ${easystack_file}..."
+                  ./check_missing_installations.sh ${easystack_file}
+                  ec=$?
+                  if [[ ${ec} -ne 0 ]]; then echo "missing installations found for ${easystack_file}!" >&2; exit ${ec}; fi
+              done
 
         - name: Test check_missing_installations.sh with missing package (GCC/8.3.0)
-          if: '!cancelled()'
           run: |
+              export EESSI_SOFTWARE_SUBDIR_OVERRIDE=${{matrix.EESSI_SOFTWARE_SUBDIR_OVERRIDE}}
               source /cvmfs/pilot.nessi.no/versions/${{matrix.EESSI_VERSION}}/init/bash
+              # set $EESSI_CPU_FAMILY to the CPU architecture that corresponds to $EESSI_SOFTWARE_SUBDIR_OVERRIDE (part before the first slash),
+              # to prevent issues with checks in the Easybuild configuration that use this variable
+              export EESSI_CPU_FAMILY=${EESSI_SOFTWARE_SUBDIR_OVERRIDE%%/*}
               module load EasyBuild
               which eb
               eb --version
               export EESSI_PREFIX=/cvmfs/pilot.nessi.no/versions/${{matrix.EESSI_VERSION}}
               export EESSI_OS_TYPE=linux
-              export EESSI_SOFTWARE_SUBDIR=${{matrix.EESSI_SOFTWARE_SUBDIR}}
               env | grep ^EESSI | sort
               # create dummy easystack file with a single entry (something that is not installed in EESSI)
               easystack_file="test.yml"

.github/workflows/test_eessi_container_script.yml

@@ -45,7 +45,7 @@ jobs:
               elif [[ ${{matrix.SCRIPT_TEST}} == 'listrepos_default' ]]; then
                 outfile=out_listrepos.txt
                 ./eessi_container.sh --verbose --list-repos | tee ${outfile}
-                grep "EESSI-pilot" ${outfile}
+                grep "EESSI" ${outfile}
 
               # test use of --list-repos with custom repos.cfg
               elif [[ ${{matrix.SCRIPT_TEST}} == 'listrepos_custom' ]]; then
@@ -57,7 +57,7 @@ jobs:
                 echo "[EESSI/20HT.TP]" >> cfg/repos.cfg
                 echo "repo_version = 20HT.TP" >> cfg/repos.cfg
                 ./eessi_container.sh --verbose --list-repos | tee ${outfile}
-                grep "EESSI-pilot" ${outfile}
+                grep "EESSI" ${outfile}
 
                 export EESSI_REPOS_CFG_DIR_OVERRIDE=${PWD}/cfg
                 ./eessi_container.sh --verbose --list-repos | tee ${outfile2}
@@ -90,15 +90,15 @@ jobs:
               elif [[ ${{matrix.SCRIPT_TEST}} == 'readwrite' ]]; then
                 outfile=out_readwrite.txt
                 fn="test_${RANDOM}.txt"
-                echo "touch /cvmfs/pilot.eessi-hpc.org/${fn}" > test_script.sh
+                echo "touch /cvmfs/pilot.nessi.no/${fn}" > test_script.sh
                 chmod u+x test_script.sh
                 export SINGULARITY_BIND="$PWD:/test"
                 ./eessi_container.sh --verbose --access rw --mode run /test/test_script.sh > ${outfile}
 
                 tmpdir=$(grep "\-\-resume" ${outfile} | sed "s/.*--resume \([^']*\).*/\1/g")
                 # note: must use '--access rw' again here, since touched file is in overlay upper dir
-                ./eessi_container.sh --verbose --resume ${tmpdir} --access rw --mode shell <<< "ls -l /cvmfs/pilot.eessi-hpc.org/${fn}" > ${outfile}
-                grep "/cvmfs/pilot.eessi-hpc.org/${fn}$" $outfile
+                ./eessi_container.sh --verbose --resume ${tmpdir} --access rw --mode shell <<< "ls -l /cvmfs/pilot.nessi.no/${fn}" > ${outfile}
+                grep "/cvmfs/pilot.nessi.no/${fn}$" $outfile
 
               # test use of --resume
               elif [[ ${{matrix.SCRIPT_TEST}} == 'resume' ]]; then
@@ -120,12 +120,12 @@ jobs:
               elif [[ ${{matrix.SCRIPT_TEST}} == 'save' ]]; then
                 outfile=out_save.txt
                 fn="test_${RANDOM}.txt"
-                test_cmd="touch /cvmfs/pilot.eessi-hpc.org/${fn}"
+                test_cmd="touch /cvmfs/pilot.nessi.no/${fn}"
                 ./eessi_container.sh --verbose --mode shell --access rw --save test-save.tar <<< "${test_cmd}" 2>&1 | tee ${outfile}
                 rm -f ${outfile}
 
-                ./eessi_container.sh --verbose --mode shell --access rw --resume test-save.tar <<< "ls -l /cvmfs/pilot.eessi-hpc.org/${fn}" > ${outfile}
-                grep "/cvmfs/pilot.eessi-hpc.org/${fn}$" $outfile
+                ./eessi_container.sh --verbose --mode shell --access rw --resume test-save.tar <<< "ls -l /cvmfs/pilot.nessi.no/${fn}" > ${outfile}
+                grep "/cvmfs/pilot.nessi.no/${fn}$" $outfile
 
                 tar tfv test-save.tar | grep "overlay-upper/${fn}"
 

.github/workflows/test_licenses.yml

@@ -0,0 +1,20 @@
+# documentation: https://help.github.com/en/articles/workflow-syntax-for-github-actions
+name: Test software licenses
+on: [push, pull_request]
+permissions:
+  contents: read # to fetch code (actions/checkout)
+jobs:
+  build:
+    runs-on: ubuntu-20.04
+    steps:
+        - name: Check out software-layer repository
+          uses: actions/checkout@93ea575cb5d8a053eaa0ac8fa3b40d7e05a33cc8 # v3.1.0
+
+        - name: set up Python
+          uses: actions/setup-python@13ae5bb136fac2878aff31522b9efb785519f984 # v4.3.0
+          with:
+            python-version: '3.9'
+
+        - name: Check software licenses
+          run: |
+            python licenses/spdx.py licenses/licenses.json

.github/workflows/tests_archdetect.yml

@@ -13,24 +13,59 @@ jobs:
           - x86_64/intel/skylake_avx512/archspec-linux-6132
           - x86_64/amd/zen2/Azure-CentOS7-7V12
           - x86_64/amd/zen3/Azure-CentOS7-7V73X
-          - ppc64le/power9le/unknown-power9le
-          - aarch64/arm/neoverse-n1/Azure-Ubuntu20-Altra
-          - aarch64/arm/neoverse-n1/AWS-awslinux-graviton2
-          - aarch64/arm/neoverse-v1/AWS-awslinux-graviton3
+          # commented out since these targets are currently not supported in pilot.nessi.no repo
+          # (and some tests assume that the corresponding subdirectory in software layer is there)
+          # - ppc64le/power9le/unknown-power9le
+          # - aarch64/neoverse-n1/Azure-Ubuntu20-Altra
+          # - aarch64/neoverse-n1/AWS-awslinux-graviton2
+          # - aarch64/neoverse-v1/AWS-awslinux-graviton3
       fail-fast: false
     steps:
     - name: checkout
       uses: actions/checkout@93ea575cb5d8a053eaa0ac8fa3b40d7e05a33cc8 # v3.1.0
 
+    - name: Mount NESSI CernVM-FS repository
+      uses: cvmfs-contrib/github-action-cvmfs@d4641d0d591c9a5c3be23835ced2fb648b44c04b # v3.1
+      with:
+          cvmfs_config_package: https://github.com/NorESSI/filesystem-layer/releases/download/latest/cvmfs-config-nessi_latest_all.deb
+          cvmfs_http_proxy: DIRECT
+          cvmfs_repositories: pilot.nessi.no
+
     - name: test eessi_archdetect.sh
       run: |
           export EESSI_MACHINE_TYPE=${{matrix.proc_cpuinfo}}
           export EESSI_MACHINE_TYPE=${EESSI_MACHINE_TYPE%%/*}
           export EESSI_PROC_CPUINFO=./tests/archdetect/${{matrix.proc_cpuinfo}}.cpuinfo
+          # check that printing of best match works correctly
           CPU_ARCH=$(./init/eessi_archdetect.sh cpupath)
           if [[ $CPU_ARCH == "$( cat ./tests/archdetect/${{matrix.proc_cpuinfo}}.output )" ]]; then
-              echo "Test for ${{matrix.proc_cpuinfo}} PASSED: $CPU_ARCH" >&2
+              echo "Test for ${{matrix.proc_cpuinfo}} PASSED: $CPU_ARCH"
           else
               echo "Test for ${{matrix.proc_cpuinfo}} FAILED: $CPU_ARCH" >&2
               exit 1
           fi
+          # check that $EESSI_SOFTWARE_SUBDIR_OVERRIDE is honored
+          export EESSI_SOFTWARE_SUBDIR_OVERRIDE='dummy/cpu'
+          CPU_ARCH=$(./init/eessi_archdetect.sh cpupath)
+          if [[ $CPU_ARCH == "${EESSI_SOFTWARE_SUBDIR_OVERRIDE}" ]]; then
+              echo "Test for picking up on \$EESSI_SOFTWARE_SUBDIR_OVERRIDE PASSED"
+          else
+              echo "Test for picking up on \$EESSI_SOFTWARE_SUBDIR_OVERRIDE FAILED" >&2
+              exit 1
+          fi
+          unset EESSI_SOFTWARE_SUBDIR_OVERRIDE
+          # check that printing of all matches works correctly (-a option for cpupath action)
+          CPU_ARCHES=$(./init/eessi_archdetect.sh -a cpupath)
+          if [[ $CPU_ARCHES == "$( cat ./tests/archdetect/${{matrix.proc_cpuinfo}}.all.output )" ]]; then
+              echo "Test for ${{matrix.proc_cpuinfo}} PASSED: $CPU_ARCHES"
+          else
+              echo "Test for ${{matrix.proc_cpuinfo}} FAILED: $CPU_ARCHES" >&2
+              exit 1
+          fi
+          # Check all those architectures actually exist (if this EESSI version has been populated already)
+          if [ -d ${EESSI_PREFIX}/software/linux ]; then
+            for dir in $(echo "$CPU_ARCHES" | tr ':' '\n'); do
+                # Search all EESSI versions as we may drop support at some point
+                ls -d ${EESSI_PREFIX}/software/linux/${dir}
+            done
+          fi

.github/workflows/tests_scripts.yml

@@ -5,7 +5,7 @@ on:
     paths:
       - build_container.sh
       - create_directory_tarballs.sh
-      - EESSI-pilot-install-software.sh
+      - EESSI-install-software.sh
       - install_software_layer.sh
       - load_easybuild_module.sh
       - run_in_compat_layer_env.sh
@@ -16,7 +16,7 @@ on:
     paths:
       - build_container.sh
       - create_directory_tarballs.sh
-      - EESSI-pilot-install-software.sh
+      - EESSI-install-software.sh
       - install_software_layer.sh
       - load_easybuild_module.sh
       - run_in_compat_layer_env.sh
@@ -40,7 +40,7 @@ jobs:
         # bind current directory into container as /software-layer
         export SINGULARITY_BIND="${PWD}:/software-layer"
 
-        # can't test with EasyBuild versions older than v4.5.2 when using EESSI pilot 2023.06,
+        # can't test with EasyBuild versions older than v4.5.2 when using EESSI 2023.06,
         # since Python in compat layer is Python 3.11.x;
         # testing with a single EasyBuild version takes a while in GitHub Actions, so stick to a single sensible version
         for EB_VERSION in '4.6.0'; do

EESSI-install-software.sh

@@ -187,31 +187,46 @@ fi
 # assume there's only one diff file that corresponds to the PR patch file
 pr_diff=$(ls [0-9]*.diff | head -1)
 
+# install any additional required scripts
+# order is important: these are needed to install a full CUDA SDK in host_injections
+# for now, this just reinstalls all scripts. Note the most elegant, but works
+${TOPDIR}/install_scripts.sh --prefix ${EESSI_PREFIX}
+
+# Install full CUDA SDK in host_injections
+# Hardcode this for now, see if it works
+# TODO: We should make a nice yaml and loop over all CUDA versions in that yaml to figure out what to install
+${EESSI_PREFIX}/scripts/gpu_support/nvidia/install_cuda_host_injections.sh -c 12.1.1 --accept-cuda-eula
+
+# Install drivers in host_injections
+# TODO: this is commented out for now, because the script assumes that nvidia-smi is available and works;
+#       if not, an error is produced, and the bot flags the whole build as failed (even when not installing GPU software)
+# ${EESSI_PREFIX}/scripts/gpu_support/nvidia/link_nvidia_host_libraries.sh
+
 # use PR patch file to determine in which easystack files stuff was added
 changed_easystacks=$(cat ${pr_diff} | grep '^+++' | cut -f2 -d' ' | sed 's@^[a-z]/@@g' | grep '^easystacks/.*yml$' | egrep -v 'known-issues|missing') 
 if [ -z ${changed_easystacks} ]; then
     echo "No missing installations, party time!"  # Ensure the bot report success, as there was nothing to be build here
 else
     for easystack_file in ${changed_easystacks}; do
-
+    
         echo -e "Processing easystack file ${easystack_file}...\n\n"
-
+    
         # determine version of EasyBuild module to load based on EasyBuild version included in name of easystack file
         eb_version=$(echo ${easystack_file} | sed 's/.*eb-\([0-9.]*\).*/\1/g')
-
+    
         # load EasyBuild module (will be installed if it's not available yet)
         source ${TOPDIR}/load_easybuild_module.sh ${eb_version}
-
+    
         ${EB} --show-config
-
+    
         echo_green "All set, let's start installing some software with EasyBuild v${eb_version} in ${EASYBUILD_INSTALLPATH}..."
-
+    
         if [ -f ${easystack_file} ]; then
             echo_green "Feeding easystack file ${easystack_file} to EasyBuild..."
-
+    
             ${EB} --easystack ${TOPDIR}/${easystack_file} --robot
             ec=$?
-
+    
             # copy EasyBuild log file if EasyBuild exited with an error
             if [ ${ec} -ne 0 ]; then
                 eb_last_log=$(unset EB_VERBOSE; eb --last-log)
@@ -221,18 +236,21 @@ else
                 # copy to build logs dir (with context added)
                 copy_build_log "${eb_last_log}" "${build_logs_dir}"
             fi
-
+    
             $TOPDIR/check_missing_installations.sh ${TOPDIR}/${easystack_file}
         else
             fatal_error "Easystack file ${easystack_file} not found!"
         fi
-
+    
     done
 fi
 
+### add packages here
+
 echo ">> Creating/updating Lmod cache..."
 export LMOD_RC="${EASYBUILD_INSTALLPATH}/.lmod/lmodrc.lua"
-if [ ! -f $LMOD_RC ]; then
+lmodrc_changed=$(cat ${pr_diff} | grep '^+++' | cut -f2 -d' ' | sed 's@^[a-z]/@@g' | grep '^create_lmodrc.py$' > /dev/null; echo $?)
+if [ ! -f $LMOD_RC ] || [ ${lmodrc_changed} == '0' ]; then
     python3 $TOPDIR/create_lmodrc.py ${EASYBUILD_INSTALLPATH}
     check_exit_code $? "$LMOD_RC created" "Failed to create $LMOD_RC"
 fi

bot/build.sh

@@ -176,6 +176,11 @@ mkdir -p ${TARBALL_TMP_BUILD_STEP_DIR}
 declare -a BUILD_STEP_ARGS=()
 BUILD_STEP_ARGS+=("--save" "${TARBALL_TMP_BUILD_STEP_DIR}")
 BUILD_STEP_ARGS+=("--storage" "${STORAGE}")
+# add options required to handle NVIDIA support
+BUILD_STEP_ARGS+=("--nvidia" "all")
+if [[ ! -z ${SHARED_FS_PATH} ]]; then
+    BUILD_STEP_ARGS+=("--host-injections" "${SHARED_FS_PATH}/host-injections")
+fi
 
 # prepare arguments to install_software_layer.sh (specific to build step)
 declare -a INSTALL_SCRIPT_ARGS=()

bot/check-result.sh

@@ -0,0 +1 @@
+check-build.sh

bot/check-test.sh

@@ -0,0 +1,20 @@
+#!/bin/bash
+#
+# Dummy script that only creates test result file for the bot, without actually checking anything
+#
+# This script is part of the EESSI software layer, see
+# https://github.com/EESSI/software-layer.git
+#
+# author: Kenneth Hoste (HPC-UGent)
+#
+# license: GPLv2
+#
+job_dir=${PWD}
+job_out="slurm-${SLURM_JOB_ID}.out"
+job_test_result_file="_bot_job${SLURM_JOB_ID}.test"
+
+echo "[TEST]" > ${job_test_result_file}
+echo "comment_description = <em>(no tests yet)</em>" >> ${job_test_result_file}
+echo "status = SUCCESS" >> ${job_test_result_file}
+
+exit 0