Merge branch '2023.06-software.eessi.io' into 2023.06-software.eessi.…

…io_Paraver

.github/workflows/modules/fake_module.lua

@@ -0,0 +1,3 @@
+setenv("INSIDE_GITHUB_ACTIONS", "true")
+-- Interfere with PATH so Lmod keeps a record
+prepend_path("PATH", "/snap/bin")

.github/workflows/scripts/only_latest_easystacks.sh

@@ -0,0 +1,58 @@
+#!/bin/bash
+#
+# This script figures out the latest version of EasyBuild being used for the installation of easystack
+# files.
+#
+# This file is part of the EESSI software layer, see
+# https://github.com/EESSI/software-layer.git
+#
+# author: Alan O'Cais (CECAM)
+#
+# license: GPLv2
+#
+
+EESSI_VERSION=${EESSI_VERSION:-"2023.06"}
+
+directory="easystacks/software.eessi.io/${EESSI_VERSION}"
+# List of example filenames
+files=($(find "$directory" -name "*.yml" | grep -e '-eb-'))
+[ -n "$DEBUG" ] && echo "${files[@]}"
+
+versions=()
+# Loop over each filename
+for filename in "${files[@]}"; do
+    # Extract the semantic version using grep
+    version=$(echo "$filename" | grep -oP '(?<=eb-)\d+\.\d+\.\d+?(?=-)')
+
+    # Output the result
+    [ -n "$DEBUG" ] && echo "Filename: $filename"
+    [ -n "$DEBUG" ] && echo "Extracted version: $version"
+    [ -n "$DEBUG" ] && echo
+    versions+=("$version")
+done
+highest_version=$(printf "%s\n" "${versions[@]}" | sort -V | tail -n 1)
+
+[ -n "$DEBUG" ] && echo "Highest version: $highest_version"
+[ -n "$DEBUG" ] && echo
+[ -n "$DEBUG" ] && echo "Matching files:"
+all_latest_easystacks=($(find $directory -type f -name "*eb-$highest_version*.yml"))
+
+accel_latest_easystacks=()
+cpu_latest_easystacks=()
+
+# Loop through the array and split based on partial matching of string
+accel="/accel/"
+for item in "${all_latest_easystacks[@]}"; do
+  if [[ "$item" == *"$accel"* ]]; then
+    accel_latest_easystacks+=("$item")
+  else
+    cpu_latest_easystacks+=("$item")
+  fi
+done
+
+# Output the results
+if [ -n "$ACCEL_EASYSTACKS" ]; then
+    echo "${accel_latest_easystacks[@]}"
+else
+    echo "${cpu_latest_easystacks[@]}"
+fi

.github/workflows/scripts/test_init_scripts.sh

@@ -0,0 +1,50 @@
+#!/bin/bash
+EESSI_VERSION="2023.06"
+export LMOD_PAGER=cat
+
+# initialize assert framework
+if [ ! -d assert.sh ]; then
+	echo "assert.sh not cloned."
+	echo ""
+	echo "run \`git clone https://github.com/lehmannro/assert.sh.git\`"
+	exit 1
+fi
+. assert.sh/assert.sh
+
+TEST_SHELLS=("bash" "zsh" "fish" "ksh")
+SHELLS=$@
+
+for shell in ${SHELLS[@]}; do
+	echo = | awk 'NF += (OFS = $_) + 100'
+	echo  RUNNING TESTS FOR SHELL: $shell
+	echo = | awk 'NF += (OFS = $_) + 100'
+  if [[ ! " ${TEST_SHELLS[*]} " =~ [[:space:]]${shell}[[:space:]] ]]; then
+		### EXCEPTION FOR CSH ###
+		echo -e "\033[33mWe don't now how to test the shell '$shell', PRs are Welcome.\033[0m" 
+  else
+		# TEST 1: Source Script and check Module Output
+		assert "$shell -c 'source init/lmod/$shell' 2>&1 " "EESSI/$EESSI_VERSION loaded successfully"
+		# TEST 2: Check if module overviews first section is the loaded EESSI module
+		MODULE_SECTIONS=($($shell -c "source init/lmod/$shell 2>/dev/null; module ov 2>&1 | grep -e '---'"))
+		PATTERN="/cvmfs/software\.eessi\.io/versions/$EESSI_VERSION/software/linux/x86_64/(intel/haswell|amd/zen3)/modules/all"
+		assert_raises 'echo "${MODULE_SECTIONS[1]}" | grep -E "$PATTERN"'
+		# TEST 3: Check if module overviews second section is the EESSI init module
+		assert "echo ${MODULE_SECTIONS[4]}" "/cvmfs/software.eessi.io/versions/$EESSI_VERSION/init/modules"
+		# Test 4: Load Python module and check version
+		command="$shell -c 'source init/lmod/$shell 2>/dev/null; module load Python/3.10.8-GCCcore-12.2.0; python --version'"
+		expected="Python 3.10.8"
+		assert "$command" "$expected"
+		# Test 5: Load Python module and check path
+		PYTHON_PATH=$($shell -c "source init/lmod/$shell 2>/dev/null; module load Python/3.10.8-GCCcore-12.2.0; which python")
+		PATTERN="/cvmfs/software\.eessi\.io/versions/$EESSI_VERSION/software/linux/x86_64/(intel/haswell|amd/zen3)/software/Python/3\.10\.8-GCCcore-12\.2\.0/bin/python"
+		echo "$PYTHON_PATH" | grep -E "$PATTERN"
+		assert_raises 'echo "$PYTHON_PATH" | grep -E "$PATTERN"'
+
+		#End Test Suite
+		assert_end "source_eessi_$shell"
+	fi
+done
+
+
+# RESET PAGER
+export LMOD_PAGER=

.github/workflows/test-software.eessi.io.yml

@@ -7,6 +7,12 @@ on:
   workflow_dispatch:
 permissions:
   contents: read # to fetch code (actions/checkout)
+env:
+  EESSI_ACCELERATOR_TARGETS: |
+    x86_64/amd/zen2:
+      - nvidia/cc80
+    x86_64/amd/zen3:
+      - nvidia/cc80
 jobs:
   check_missing:
     runs-on: ubuntu-latest
@@ -21,6 +27,7 @@ jobs:
         - aarch64/neoverse_v1
         - x86_64/amd/zen2
         - x86_64/amd/zen3
+        - x86_64/amd/zen4
         - x86_64/intel/haswell
         - x86_64/intel/skylake_avx512
         - x86_64/generic
@@ -48,14 +55,48 @@ jobs:
               export EESSI_PREFIX=/cvmfs/software.eessi.io/versions/${{matrix.EESSI_VERSION}}
               export EESSI_OS_TYPE=linux
               env | grep ^EESSI | sort
-              echo "just run check_missing_installations.sh (should use easystacks/software.eessi.io/${{matrix.EESSI_VERSION}}/eessi-${{matrix.EESSI_VERSION}}-*.yml)"
-              for easystack_file in $(ls easystacks/software.eessi.io/${{matrix.EESSI_VERSION}}/eessi-${{matrix.EESSI_VERSION}}-eb-*.yml); do
+
+              # first check the CPU-only builds for this CPU target
+              echo "just run check_missing_installations.sh (should use easystacks/software.eessi.io/${{matrix.EESSI_VERSION}}/eessi-${{matrix.EESSI_VERSION}}-*.yml with latest EasyBuild release)"
+              for easystack_file in $(EESSI_VERSION=${{matrix.EESSI_VERSION}} .github/workflows/scripts/only_latest_easystacks.sh); do
+                  if [ ${{matrix.EESSI_SOFTWARE_SUBDIR_OVERRIDE}} = "x86_64/amd/zen4" ]; then
+                      if grep -q 2022b <<<"${easystack_file}"; then
+                        # skip the check of installed software on zen4 for foss/2022b builds
+                        continue
+                      fi
+                      if [[ $easystack_file == *"rebuilds"* ]]; then
+                        # Also handle rebuilds, make a temporary EasyStack file where we clean out all 2022b stuff and use that
+                        new_easystack=$(mktemp pruned_easystackXXX --suffix=.yml)
+                        # first clean out the options then clean out the .eb name
+                        sed '/2022b\|12\.2\.0/,/\.eb/{/\.eb/!d}' "${easystack_file}" | sed '/2022b\|12\.2\.0/d' > $new_easystack
+                        diff --unified=0 "$easystack_file" "$new_easystack" || :
+                        easystack_file="$new_easystack"
+                      fi
+                  fi
                   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
 
+              # now check the accelerator builds for this CPU target
+              accelerators=$(echo "${EESSI_ACCELERATOR_TARGETS}" | yq ".${EESSI_SOFTWARE_SUBDIR_OVERRIDE}[]")
+              if [ -z ${accelerators} ]; then
+                  echo "no accelerator targets defined for ${EESSI_SOFTWARE_SUBDIR_OVERRIDE}"
+              else
+                  for accel in ${accelerators}; do
+                      module use ${EESSI_SOFTWARE_PATH}/accel/${accel}/modules/all
+                      echo "checking missing installations for accelerator ${accel} using modulepath: ${MODULEPATH}"
+                      for easystack_file in $(EESSI_VERSION=${{matrix.EESSI_VERSION}} ACCEL_EASYSTACKS=1 .github/workflows/scripts/only_latest_easystacks.sh); 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
+                      module unuse ${EESSI_SOFTWARE_PATH}/accel/${accel}/modules/all
+                  done
+              fi
+
         - name: Test check_missing_installations.sh with missing package (GCC/8.3.0)
           run: |
               export EESSI_SOFTWARE_SUBDIR_OVERRIDE=${{matrix.EESSI_SOFTWARE_SUBDIR_OVERRIDE}}

.github/workflows/tests_archdetect_nvidia_gpu.yml

@@ -0,0 +1,124 @@
+# documentation: https://help.github.com/en/articles/workflow-syntax-for-github-actions
+name: Tests for accelerator detection (NVIDIA GPU)
+on:
+  push:
+  pull_request:
+permissions:
+  contents: read # to fetch code (actions/checkout)
+jobs:
+  build:
+    runs-on: ubuntu-latest
+    strategy:
+      matrix:
+        fake_nvidia_smi_script:
+          - none # no nvidia-smi command
+          - no_devices # nvidia-smi command works, but no GPUs available
+          - 1xa100 # cc80, supported with (atleast) zen2 CPU
+          - 2xa100 # cc80, supported with (atleast) zen2 CPU
+          - 4xa100 # cc80, supported with (atleast) zen2 CPU
+          - cc01 # non-existing GPU
+      fail-fast: false
+    steps:
+    - name: checkout
+      uses: actions/checkout@d632683dd7b4114ad314bca15554477dd762a938 # v4.2.0
+
+    # we deliberately do not use the eessi/github-action-eessi action,
+    # because we want to control when the EESSI environment is initialized
+    - name: Mount EESSI CernVM-FS repository
+      uses: cvmfs-contrib/github-action-cvmfs@55899ca74cf78ab874bdf47f5a804e47c198743c # v4.0
+      with:
+          cvmfs_config_package: https://github.com/EESSI/filesystem-layer/releases/download/latest/cvmfs-config-eessi_latest_all.deb
+          cvmfs_http_proxy: DIRECT
+          cvmfs_repositories: software.eessi.io
+
+    - name: test accelerator detection
+      run: |
+          export EESSI_SOFTWARE_SUBDIR_OVERRIDE='x86_64/amd/zen2'
+
+          # put fake nvidia-smi command in place (unless we don't want to)
+          if [[ "${{matrix.fake_nvidia_smi_script}}" != "none" ]]; then
+              tmpdir=$(mktemp -d)
+              ln -s $PWD/tests/archdetect/nvidia-smi/${{matrix.fake_nvidia_smi_script}}.sh $tmpdir/nvidia-smi
+              export PATH=$tmpdir:$PATH
+          fi
+
+          # first run with debugging enabled, just to show the output
+          ./init/eessi_archdetect.sh -d accelpath || echo "non-zero exit code: $?"
+
+          # verify output (or exit code if non-zero)
+          out=$(./init/eessi_archdetect.sh accelpath || echo "non-zero exit code: $?")
+
+          if [[ $out == "$( cat ./tests/archdetect/nvidia-smi/${{matrix.fake_nvidia_smi_script}}.output )" ]]; then
+
+              echo "Test for '${{matrix.fake_nvidia_smi_script}}' PASSED: '$out'"
+
+              # run full EESSI init script, which pick up on the accelerator (if available)
+              echo
+              . init/bash 2>&1 | tee init.out
+              echo "-----------------------------------------------------------------------------"
+
+              if [[ "${{matrix.fake_nvidia_smi_script}}" == "none" ]] || [[ "${{matrix.fake_nvidia_smi_script}}" == "no_devices" ]]; then
+
+                  pattern="archdetect could not detect any accelerators"
+                  echo ">>> checking for pattern '${pattern}' in init output..."
+                  grep "${pattern}" init.out || (echo "FAILED 1" || exit 1)
+
+                  pattern="archdetect found supported accelerator"
+                  echo ">>> checking for lack of pattern '${pattern}' in init output..."
+                  match=$(grep "${pattern}" init.out || true)
+                  test "x${match}" = "x" || (echo "unexpected match found for '${pattern}' in init output" && exit 1)
+
+                  pattern="Prepending /cvmfs/software.eessi.io/versions/2023.06/software/linux/.*/accel/.*/modules/all to \$MODULEPATH"
+                  echo ">>> checking for lack of pattern '${pattern}' in init output..."
+                  match=$(grep "${pattern}" init.out || true)
+                  test "x${match}" = "x" || (echo "unexpected match found for '${pattern}' in init output" && exit 1)
+
+              elif [[ "${{matrix.fake_nvidia_smi_script}}" == "cc01" ]]; then
+
+                  pattern="No matching path found in x86_64/amd/zen2 for accelerator detected by archdetect (accel/nvidia/cc01)"
+                  echo ">>> checking for pattern '${pattern}' in init output..."
+                  grep "${pattern}" init.out || (echo "FAILED 1" || exit 1)
+
+                  pattern="Prepending /cvmfs/software.eessi.io/versions/2023.06/software/linux/.*/accel/.*/modules/all to \$MODULEPATH"
+                  echo ">>> checking for lack of pattern '${pattern}' in init output..."
+                  match=$(grep "${pattern}" init.out || true)
+                  test "x${match}" = "x" || (echo "unexpected match found for '${pattern}' in init output" && exit 1)
+
+              else
+                  echo ">>> checking for 'accel/nvidia/cc80' in init output..."
+                  grep "archdetect found supported accelerator for CPU target x86_64/amd/zen2: accel/nvidia/cc80" init.out || (echo "FAILED 2" && exit 1)
+                  grep "Prepending /cvmfs/software.eessi.io/versions/2023.06/software/linux/x86_64/amd/zen2/accel/nvidia/cc80/modules/all to \$MODULEPATH" init.out || (echo "FAILED 3" && exit 1)
+              fi
+
+              echo ">>> checking last line of init output..."
+              tail -1 init.out | grep "Environment set up to use EESSI (2023.06), have fun!" || (echo "FAILED, full init utput:" && cat init.out && exit 1)
+
+              echo "All checks on init output PASSED"
+          else
+              echo "Test for '${{matrix.fake_nvidia_smi_script}}' FAILED: '$out'" >&2
+              exit 1
+          fi
+
+    - name: test accelerator detection under $EESSI_ACCEL_SOFTWARE_SUBDIR_OVERRIDE + $EESSI_ACCELERATOR_TARGET_OVERRIDE
+      run: |
+          export EESSI_SOFTWARE_SUBDIR_OVERRIDE='x86_64/amd/zen2'
+          export EESSI_ACCEL_SOFTWARE_SUBDIR_OVERRIDE='x86_64/amd/zen3'
+          export EESSI_ACCELERATOR_TARGET_OVERRIDE='accel/nvidia/cc80'
+
+          # first run with debugging enabled, just to show the output
+          ./init/eessi_archdetect.sh -d accelpath || echo "non-zero exit code: $?"
+
+          # verify output (or exit code if non-zero)
+          out=$(./init/eessi_archdetect.sh accelpath || echo "non-zero exit code: $?")
+
+          echo
+          . init/bash 2>&1 | tee init.out
+          echo "-----------------------------------------------------------------------------"
+
+          echo ">>> checking for 'accel/nvidia/cc80' in init output..."
+          grep "archdetect found supported accelerator for CPU target x86_64/amd/zen3: accel/nvidia/cc80" init.out || (echo "FAILED 1" && exit 1)
+          grep "Using x86_64/amd/zen2 as software subdirectory" init.out || (echo "FAILED 2" && exit 1)
+          grep "Prepending /cvmfs/software.eessi.io/versions/2023.06/software/linux/x86_64/amd/zen2/modules/all to \$MODULEPATH" init.out || (echo "FAILED 3" && exit 1)
+          grep "Prepending /cvmfs/software.eessi.io/versions/2023.06/software/linux/x86_64/amd/zen3/accel/nvidia/cc80/modules/all to \$MODULEPATH" init.out || (echo "FAILED 4" && exit 1)
+
+          echo "All checks on init output PASSED"