Script measures IO activity of the processes for a specified amount of time and return top list of those. Read, write or both IO operation could be measured and reported.

topio.py

#!/usr/bin/python

import subprocess
import sys
import time
from optparse import OptionParser

def getOptionParser():
    parser = OptionParser(version = '0.1_2012017',
                          description = '''Script measures IO activity of the processes for a specified amount of time
                          and return top list of those. Read, write or both IO operation could be measured and reported.
                          romans(dot)krjukovs(at)gmail(dot)com''')
    
    parser.add_option("-i", "--interval", 
                      dest = "time_interval",
                      help = "time interval between making proc io snapshots",
                      action = "store",
                      type = "int",
                      default = 5,
                      metavar = "SECONDS")
    
    parser.add_option("-o", "--operation", 
                      dest = "operation",
                      help = "operation to measure IO on",
                      action = "store",
                      type = "choice",
                      choices = ('R', 'W', 'RW'),
                      default = "W",
                      metavar = 'OPERATION') 
    
    parser.add_option("-v", "--verbose",
                      action = "store_true", 
                      dest = "verbose", 
                      default = False)

    parser.add_option("-n", "--no-heading",
                      action = "store_true", 
                      dest = "noheading", 
                      default = False)
    
    return parser


def run_command(command):
    p = subprocess.Popen(command,
                         stdout=subprocess.PIPE,
                         stderr=subprocess.STDOUT)
    return iter(p.stdout.readline, '')

def read_pid_io(pid):
	res = {}
	for line in run_command(["cat", "/proc/"+pid+"/io"]):
		if line.find("bytes")>-1:
			line_split = line.split(":")
			res[line_split[0]] = line_split[1].strip()
	return res

def snapshot_pid_io():
	res = {}
	for line in run_command(["ps", "-e", "--no-heading"]):
		parts = line.split()    
		res[parts[0]]={'pid':parts[0], 'proc':parts[3], 'io':read_pid_io(parts[0])}
	return res

def compare(f,s,switch):
	res = {}
	for pid,details in f.items():
		if len(details['io'].keys())>0:
			first_bytes = int(details['io'][switch])
			second_bytes = int(s[pid]['io'][switch])
			if (second_bytes > first_bytes):
				res[pid] = second_bytes - first_bytes
	return res

def parse(pid_hash, compare_res, options, operation):
    if not(options.noheading):
        print '%s %s %s' % ('PID'.rjust(5), (operation+'-Bytes').rjust(10), 'Proc'.ljust(16))

    for pid, delta in sorted(compare_res.items(), key=lambda x: x[1]):
         print '%s %s %s' % (str(pid).rjust(5), str(delta).rjust(10), str(pid_hash[pid]['proc']).ljust(16))

if __name__ == "__main__":
    
    (options, args) = getOptionParser().parse_args()
    
    if options.verbose:
        print "Making initial snapshot of proc io.."
    first = snapshot_pid_io()

    if options.verbose:
        print "Pausing for %d sec.." % (options.time_interval)
    time.sleep(options.time_interval)

    if options.verbose:
        print "Making second snapshot of proc io.."
    second = snapshot_pid_io()

    if 'R' in options.operation:
        if options.verbose:
            print "Comparing R-bytes:"
        parse(second, compare(first, second, 'read_bytes'), options, "R") 
    
    if 'W' in options.operation:
        if options.verbose:
            print "Comparing W-bytes:"
        parse(second, compare(first, second, 'write_bytes'), options, "W")