#!/usr/bin/python
#
# Script to demonstrate the usage of 'tshark' and calculating some statistics.
#
# Pass this script a .pcap[.gz] file which contains GlusterFS traffic, and the
# total of transferred bytes per GlusterFS packet will be calculated. Some
# things to keep in mind:
# - only GlusterFS packets will be used (skipping setup/teardown of TCP)
# - size from the RPC payload is used (not adding ../IP/TCP)
# - the total includes both sent and received data
# - all GlusterFS operations related to I/O are included
# - the script is kept simple and contains little error checking
#
# Author: Niels de Vos <ndevos@redhat.com>
# Date: Sat, 23 Feb 2013
#

import sys
import os

if len(sys.argv) != 2:
    print "Usage: %s <network-capture>" % sys.argv[0]
    sys.exit(1)

filename = sys.argv[1]

tshark_command = "tshark -T fields -E header=y -E separator=, -r %s" % filename

# dictionary with fieldnames and a function for convert the output
fields = {
    "frame.number" : int,
    "rpc.repframe" : int,
    "rpc.fraglen" : int,
    "rpc.auth.uid" : str }

filter = "glusterfs"

tshark_command += " -e " + " -e ".join(fields.keys())
tshark_command += " '%s'" % (filter)

# read the output from the tshark command
output = os.popen(tshark_command)

# header will contain the fields in the correct order
header = list()
# append all packets to a list
packets = list()
for line in output:
    # Bug in certain Wireshark versions causing extra line of output
    if line.startswith("libwireshark.so"):
        continue

    values = line.strip().split(",")

    if not header:
        header = values
        continue

    packet = dict()
    for (idx, val) in enumerate(values):
        # convert according to the function in fields
        if not val and fields[header[idx]] == int:
            packet[header[idx]] = 0
        else:
            packet[header[idx]] = fields[header[idx]](val)
    packets.append(packet)

# keep a list of packets that have been merged
reap = list()
# merge the reply with the call
for p in packets:
    if p["rpc.repframe"]:
        for call in packets:
            if call["frame.number"] == p["rpc.repframe"]:
                call["rpc.fraglen"] += p["rpc.fraglen"]
                reap.append(p)
                break

# remove all the replies that have been merged
for p in reap:
    packets.remove(p)

# create a list of uids and calculate the total traffic
uids = dict()
for p in packets:
    if p["rpc.auth.uid"] not in uids.keys():
        uids[p["rpc.auth.uid"]] = 0

    uids[p["rpc.auth.uid"]] += p["rpc.fraglen"]

# print a table with the results
print "%8s | UID" % ("bytes")
print "---------+-----"
for uid in uids.keys():
    print "%8d | %s" % (uids[uid], uid)