Detect software used on the cluster with the index of Mii

.github/workflows/spelling.yaml

@@ -17,4 +17,4 @@ jobs:
       with:
         check_filenames: true
         ignore_words_list: keypair
-        skip: ./locale/fr/LC_MESSAGES/django.po,./locale/fr/LC_MESSAGES/djangojs.po
+        skip: ./locale/fr/LC_MESSAGES/django.po,./locale/fr/LC_MESSAGES/djangojs.po,./mii-parser.py,./userportal/settings/30-cluster.py

mii-parser.py

@@ -0,0 +1,144 @@
+import subprocess
+import argparse
+
+# Parse the command line arguments
+parser = argparse.ArgumentParser(description='Convert mii output to regex')
+parser.add_argument('-d', '--duplicates', action='store_true', help='show duplicates')
+parser.add_argument('-p', '--pretty', action='store_true', help='pretty print')
+parser.add_argument('-o', '--output', action='store_true', help='print output for the config file')
+args = parser.parse_args()
+
+# Run the CLI command and capture the output
+output = subprocess.check_output(["mii", "list"], universal_newlines=True)
+
+softwares = {'python': set(), 'R': set()}
+
+for line in output.splitlines():
+    try:
+        soft_version = line.strip().split('/')
+        software = soft_version[0]
+        version = soft_version[1]
+    except IndexError:
+        continue
+
+    show = subprocess.check_output(["mii", "show", line], universal_newlines=True)
+    if 'empty result set' in show:
+        continue
+
+    # rename and group some softwares
+    if software in ['python3', 'python2', 'python-build-bundle', 'pip']:
+        software = 'python'
+    if software in ['rstudio-server']:
+        software = 'R'
+    if software in ['namd-ofi', 'namd-ucx', 'namd-multicore', 'namd-ucx-smp', 'namd-ofi-smp']:
+        software = 'namd'
+    if software in ['starccm-mixed']:
+        software = 'starccm'
+    if software in ['gromacs-plumed', 'gromacs-cp2k', 'gromacs-ramd', 'gromacs-colvars']:
+        software = 'gromacs'
+    if software in ['petsc-pardiso-64bits', 'petsc-pardiso', 'petsc-64bits']:
+        software = 'petsc'
+    if software in ['vtk-mpi']:
+        software = 'vtk'
+    if software in ['hdf-fortran']:
+        software = 'hdf'
+    if software in ['blast+', 'rmblast']:
+        software = 'blast'
+    if software in ['fftw-mpi']:
+        software = 'fftw'
+    if software in ['meryl-lookup', 'meryl-import']:
+        software = 'meryl'
+    if software in ['paraview-offscreen-gpu', 'paraview-offscreen']:
+        software = 'paraview'
+    if software in ['openbabel-omp']:
+        software = 'openbabel'
+    if software in ['geant4-seq', 'geant4-topasmc3.9']:
+        software = 'geant4'
+    if software in ['hdf5-mpi']:
+        software = 'hdf5'
+    if software in ['netcdf-mpi']:
+        software = 'netcdf'
+    if software in ['phylobayes-mpi']:
+        software = 'phylobayes'
+    if software in ['ansysedt']:
+        software = 'ansys'
+    if software in ['hpcspades']:
+        software = 'spades'
+    if software in ['ambertools']:
+        software = 'amber'
+    if software in ['cudacore', 'nvhpc']:
+        software = 'cuda'
+    if software in ['mysql', 'mariadb']:
+        software = 'mysql'
+    if software in ['perl-no-thread']:
+        software = 'perl'
+    if software in ['wrf-co2', 'wrf-cmaq']:
+        software = 'wrf'
+    if software in ['scotch-no-thread']:
+        software = 'scotch'
+    if software in ['cfour-mpi']:
+        software = 'cfour'
+    if software in ['chapel-multicore', 'chapel-ucx', 'chapel-ofi']:
+        software = 'chapel'
+    if software in ['metis-64idx']:
+        software = 'metis'
+    if software in ['raxml-pthreads']:
+        software = 'raxml'
+    if software in ['mafft-mpi']:
+        software = 'mafft'
+    if software in ['flexiblascore']:
+        software = 'flexiblas'
+    if software in ['ls-dyna-mpi']:
+        software = 'ls-dyna'
+
+    if software not in softwares:
+        softwares[software] = set()  # set of binaries
+
+    for binary in show.splitlines()[1:]:
+        binary = binary.strip()
+        if binary.startswith('f2py'):
+            continue
+        if 'python' in binary or binary.startswith('pip') or binary.startswith('easy_install'):
+            software = 'python'
+        if binary == 'Rscript' or binary == 'R':
+            software = 'R'
+        softwares[software].add(binary)
+
+    if software in ['openmpi', 'intelmpi', 'dpc++', 'clang', 'llvm', 'intel', 'intelmpi', 'libfabric', 'gcccore']:
+        del softwares[software]
+    # ignore some softwares that seems to repackage the whole world or have too many binaries
+    if software in ['afni', 'ansys', 'masurca', 'minc-toolkit']:
+        del softwares[software]
+
+if args.pretty:
+    for software in sorted(softwares.keys()):
+        print(software + ":")
+        for binary in sorted(softwares[software]):
+            print("  " + binary)
+        print()
+
+if args.duplicates:
+    duplicates = {}
+    for software in softwares:
+        binaries = softwares[software]
+        for binary in binaries:
+            if binary not in duplicates:
+                duplicates[binary] = set()
+            duplicates[binary].add(software)
+    print("Duplicates:")
+    for binary in duplicates:
+        if len(duplicates[binary]) > 1:
+            print(binary + ":")
+            for software in duplicates[binary]:
+                print("  " + software)
+            print()
+
+if args.output:
+    for software in sorted(softwares.keys(), key=lambda x: len(softwares[x])):
+        if len(softwares[software]) == 0:
+            continue
+        # add quotes on each binary
+        softwares[software] = ['"' + binary + '"' for binary in softwares[software]]
+        print("    ('{software}', [{bins}]),".format(
+            software=software,
+            bins=','.join(sorted(softwares[software]))))

pages/views.py

@@ -470,14 +470,25 @@ def graph_software(query_str, software_regexes, extract_path=False):
         except KeyError:
             # Somehow the metric is missing the exe label
             continue
-        for regex, name in software_regexes:
-            if re.match(regex, bin):
-                if name in software:
-                    software[name] += value
-                else:
-                    software[name] = value
-                accounted += value
-                break
+        basename = bin.split('/')[-1]
+        for name, regex in software_regexes:
+            if isinstance(regex, list):
+                # not a regex but a list of binary names
+                if basename in regex:
+                    if name in software:
+                        software[name] += value
+                    else:
+                        software[name] = value
+                    accounted += value
+                    break
+            else:
+                if re.match(regex, bin):
+                    if name in software:
+                        software[name] += value
+                    else:
+                        software[name] = value
+                    accounted += value
+                    break
         else:
             if extract_path:
                 name = "Stored in /{}".format(bin.split('/')[1])