nodeocc 1.0.2__tar.gz → 1.0.4__tar.gz

{nodeocc-1.0.2 → nodeocc-1.0.4}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: nodeocc
-Version: 1.0.2
+Version: 1.0.4
 Summary: TUI application for viewing the status of GPU allocations on a Slurm cluster
 Author: Matteo Boschini
 Maintainer: Lorenzo Bonicelli, Aniello Panariello

{nodeocc-1.0.2 → nodeocc-1.0.4}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "nodeocc"
-version = "1.0.2"
+version = "1.0.4"
 authors = [{ name = "Matteo Boschini" }]
 maintainers = [{ name = "Lorenzo Bonicelli" }, { name = "Aniello Panariello" }]
 description = "TUI application for viewing the status of GPU allocations on a Slurm cluster"

{nodeocc-1.0.2 → nodeocc-1.0.4}/src/controller/controller.py

@@ -57,6 +57,7 @@ try:
     print(END, end='')
 except Exception as e:
     print(f"{BBLUE}Could not get newest version from pip, check if {program_name} is up to date{END}")
+    newest_version = '?.?.?'
     time.sleep(5)
 
 

{nodeocc-1.0.2 → nodeocc-1.0.4}/src/model/infrastructure.py

@@ -67,7 +67,7 @@ class Infrastructure:
         self.cpu_limit_grp = cpu_limit_grp
         self.cpu_limit_stu = cpu_limit_stu
         self.cpu_limit_stugrp = cpu_limit_stugrp
-        self.prior = ['n/a', None, 'RTXA5000', 'RTX6000', 'RTX5000', '2080', 'V100', '1080', 'P100', 'K80']
+        self.prior = ['n/a', None, 'A40', 'RTXA5000', 'RTX6000', 'RTX5000', '2080', 'V100', '1080', 'P100', 'K80']
 
     @staticmethod
     def from_dict(d):

{nodeocc-1.0.2 → nodeocc-1.0.4}/src/nodeocc.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: nodeocc
-Version: 1.0.2
+Version: 1.0.4
 Summary: TUI application for viewing the status of GPU allocations on a Slurm cluster
 Author: Matteo Boschini
 Maintainer: Lorenzo Bonicelli, Aniello Panariello

{nodeocc-1.0.2 → nodeocc-1.0.4}/src/readers/slurmreader.py

@@ -1,6 +1,5 @@
-import json
 import os
-import time
+import re
 from io import StringIO
 
 import pandas as pd
@@ -152,27 +151,29 @@ def _load_joblet_meta(line):
 
 
 def _gpus_per_joblet(line):
-    if pd.isna(line['gpus_per_job']) and pd.isna(line['gpus_per_node']) and pd.isna(line['gpus_per_task']):
+    try:
+        if 'node=1' in line['TRES_ALLOC']:
+            return line['TRES_ALLOC'].split('gpu=')[-1].split(',')[0] if 'gpu=' in line['TRES_ALLOC'] else 0
+        else:
+            node = line['NODELIST']
+            node_idx = line['ALL_JOB_NODES'].split(',').index(node)
+            return int(re.findall(r'gpu:(\d+)', line['TRES_PER_NODE'])[node_idx]) if 'gpu:' in line['TRES_PER_NODE'] else 0
+    except BaseException as e:
         return 0
-    elif not pd.isna(line['gpus_per_node']):
-        return line['gpus_per_node'] * (line['NODES'] if pd.isna(line['NODELIST']) else 1)
-    else:
-        if line['ST'] == 'PD':
-            return line['gpus_per_job'] if not pd.isna(line['gpus_per_job']) else line['gpus_per_task']
-        gpus_id = line['joblet_meta_str']
-        gpus_id = sum([x.split('IDX:')[-1].split(')')[0].split(',') for x in gpus_id if 'gpu' in x], [])
-        gpus_n = len(gpus_id) + sum([len(range(int(x.split('-')[1].split()[0]) - int(x.split('-')[0]))) for x in gpus_id if '-' in x])
-        return gpus_n
 
 
 def _cpus_per_joblet(line):
-    if line['ST'] == 'PD':
-        return line['MIN_CPUS']
-    else:
-        cpus_id = line['joblet_meta_str']
-        cpus = sum([x.split('CPU_IDs=')[-1].split()[0].split(',') for x in cpus_id], [])
-        cpus_n = len(cpus) + sum([len(range(int(x.split('-')[1].split()[0]) - int(x.split('-')[0]))) for x in cpus if '-' in x])
-        return cpus_n
+    try:
+        if 'node=1' in line['TRES_ALLOC']:
+            return int(str(line['TRES_ALLOC']).split('cpu=')[-1].split(',')[0])
+        else:
+            nodes = line['NODES']
+            tot_gpu = int(str(line['TRES_ALLOC']).split('gpu=')[-1].split(',')[0]) if 'gpu=' in str(line['TRES_ALLOC']) else 0
+            tot_cpus = int(str(line['TRES_ALLOC']).split('cpu=')[-1].split(',')[0])
+            n_gpus = line['joblet_gpus']
+            return int(tot_cpus) // int(nodes) if tot_gpu == 0 else int(tot_cpus) // tot_gpu * n_gpus
+    except BaseException as e:
+        return 0
 
 
 memunits = {
@@ -193,12 +194,24 @@ def parse_mem(thing):
 
 
 def _mem_per_joblet(line):
-    if line['ST'] == 'PD':
-        return parse_mem(line['MIN_MEMORY'])
-    else:
-        mem_id = line['joblet_meta_str']
-        mem = sum([int(x.split('Mem=')[-1].split()[0]) for x in mem_id])
+    try:
+        mem = line['TRES_ALLOC'].split('mem=')[-1].split(',')[0]
+        if 'G' in mem:
+            mem = int(mem.split('G')[0]) * 1024
+        elif 'M' in mem:
+            mem = int(mem.split('M')[0])
+        else:
+            mem = int(mem)
+    except BaseException:
+        return 0
+
+    if 'node=1' in line['TRES_ALLOC']:
         return mem
+    else:
+        nodes = line['NODES']
+        tot_gpu = int(str(line['TRES_ALLOC']).split('gpu=')[-1].split(',')[0]) if 'gpu=' in str(line['TRES_ALLOC']) else 0
+        n_gpus = line['joblet_gpus']
+        return int(mem) // int(nodes) if tot_gpu == 0 else int(mem) // tot_gpu * n_gpus
 
 
 def _true_jobid(line):
@@ -257,15 +270,20 @@ def read_jobs():
     instance.timeme(f"\t- squeue and dataframe creation")
 
     squeue_df['JOBID'] = squeue_df['JOBID'].apply(str)
+    squeue_df['ALL_JOB_NODES'] = squeue_df['NODELIST']
     squeue_df = _split_column(squeue_df, 'NODELIST')
+
     squeue_df['gpus_per_node'] = squeue_df['TRES_PER_NODE'].apply(lambda x: try_parse_gpu(x) if isinstance(x, str) else x)
     squeue_df['gpus_per_job'] = squeue_df['TRES_PER_JOB'].apply(lambda x: try_parse_gpu(x) if isinstance(x, str) else x)
     squeue_df['gpus_per_task'] = squeue_df['TRES_PER_TASK'].apply(lambda x: try_parse_gpu(x) if isinstance(x, str) else x)
 
     try:
-        squeue_df['joblet_mem'] = squeue_df['TRES_ALLOC'].apply(get_mem_joblet)
-        squeue_df['joblet_gpus'] = squeue_df['TRES_ALLOC'].apply(lambda l: str(l).split('gpu=')[-1].split(',')[0] if 'gpu=' in str(l) else 0)
-        squeue_df['joblet_cpus'] = squeue_df['TRES_ALLOC'].apply(lambda l: int(str(l).split('cpu=')[-1].split(',')[0]))
+        # squeue_df['joblet_gpus'] = squeue_df['TRES_ALLOC'].apply(lambda l: str(l).split('gpu=')[-1].split(',')[0] if 'gpu=' in str(l) else 0)
+        squeue_df['joblet_gpus'] = squeue_df.apply(_gpus_per_joblet, axis=1)
+        # squeue_df['joblet_cpus'] = squeue_df['TRES_ALLOC'].apply(lambda l: int(str(l).split('cpu=')[-1].split(',')[0]))
+        squeue_df['joblet_cpus'] = squeue_df.apply(_cpus_per_joblet, axis=1)
+        # squeue_df['joblet_mem'] = squeue_df['TRES_ALLOC'].apply(get_mem_joblet)
+        squeue_df['joblet_mem'] = squeue_df.apply(_mem_per_joblet, axis=1)
     except Exception as e:
         instance.err(f"Exception PD: {e}")
 

{nodeocc-1.0.2 → nodeocc-1.0.4}/src/view/curses_multiwindow.py

@@ -89,7 +89,7 @@ class Singleton:
         self.basepath = self.args.basepath
 
         if self.args.daemon_only:
-            handler = RotatingFileHandler(os.path.join(self.basepath, '.master_log_n.txt'), maxBytes=5 * 1024 * 1024, backupCount=2, mode='w')
+            handler = RotatingFileHandler(os.path.join(self.basepath, '.master_log.txt'), maxBytes=5 * 1024 * 1024, backupCount=2, mode='w')
             logging.basicConfig(level=logging.DEBUG, format='%(asctime)s - %(message)s', datefmt='%d-%b-%y %H:%M:%S',
                                 handlers=[handler])
         # set logging file

{nodeocc-1.0.2 → nodeocc-1.0.4}/src/view/slurm_list.py

@@ -47,7 +47,7 @@ def _joblet_format(instance, job, width=74, jobid_type='agg'):
         if instance.view_mode in ('gpu', 'ram'):
             joblet_repr += _format_to((f'{joblet.n_gpus}gp' if joblet.n_gpus > 0 else " - "), 3, 'left')
         elif instance.view_mode == 'cpu':
-            joblet_repr += _format_to((f'{joblet.cpus}cp' if joblet.cpus > 0 else "-"), 3, 'left')
+            joblet_repr += _format_to((f'{joblet.cpus}cp' if joblet.cpus > 0 else " - "), 3, 'left')
         joblet_repr += ' '
         joblet_repr += _format_to(joblet.node if joblet.node is not None else job.reason, 11)
         joblet_reprs.append(joblet_repr)