pdsys

Pandas-powered package for systems monitoring

To Inatall

pip install pdsys

Basic usage

import pdsys

To get a system utilization report (as dataframe) on local:

df = pdsys.report()

pdsys returns by default all process attributes information from psutil process iterator:

df.columns.tolist()
>>> ['cmdline',
 'connections',
 'cpu_affinity',
 'cpu_num',
 'cpu_percent',
 'cpu_times.children_system',
 'cpu_times.children_user',
 'cpu_times.system',
 'cpu_times.user',
 'create_time',
 'cwd',
 'environ',
 'exe',
 'gids.effective',
 'gids.real',
 'gids.saved',
 'hostname',
 'io_counters',
 'io_counters.read_bytes',
 'io_counters.read_chars',
 'io_counters.read_count',
 'io_counters.write_bytes',
 'io_counters.write_chars',
 'io_counters.write_count',
 'ionice.value',
 'memory_full_info',
 'memory_full_info.data',
 'memory_full_info.dirty',
 'memory_full_info.lib',
 'memory_full_info.pss',
 'memory_full_info.rss',
 'memory_full_info.shared',
 'memory_full_info.swap',
 'memory_full_info.text',
 'memory_full_info.uss',
 'memory_full_info.vms',
 'memory_info.data',
 'memory_info.dirty',
 'memory_info.lib',
 'memory_info.rss',
 'memory_info.shared',
 'memory_info.text',
 'memory_info.vms',
 'memory_maps',
 'memory_percent',
 'name',
 'nice',
 'num_ctx_switches.involuntary',
 'num_ctx_switches.voluntary',
 'num_fds',
 'num_threads',
 'open_files',
 'pid',
 'ppid',
 'status',
 'terminal',
 'threads',
 'uids.effective',
 'uids.real',
 'uids.saved',
 'username']

You can query the output dataframe to get more insights about the system:

# getting top 5 processes sorted by memory utilization
df.sort_values(by='memory_percent',
               ascending=False)[['name', 'memory_percent']].head(5)

Also, pdsys can run reports from remote systems by providing list of hosts:

df = pdsys.report(hosts=['user@host1', 'user@host2'])
df[df.memory_percent > 0.9].groupby(['hostname',
                                     'name']).agg({'memory_percent': 'sum',
                                                   'pid': 'count',
                                                   'num_threads': 'sum',
                                                   'memory_info.rss': lambda x: sum(x) / 1e6})