1

from rrdweb import rrd

     2

     3

import operator

     4

     5

def host_def (idx, name, dir, ds_in, ds_out, cf) :

     6

    """

     7

        VDEFs for given host, giving its avg/max values:

     8

     9

            DEF:in/out_{idx}

    10

            CDEFtraf_{idx}

    11

            VDEF:avg/max_{idx}

    12

            PRINT:{idx} avg/max {value}

    13

    """

    14

    15

    params = dict(

    16

        dir=dir, ds_in=ds_in, ds_out=ds_out, cf=cf,

    17

        idx=idx, rrd='%s/%s.rrd' % (dir, name)

    18

    )

    19

    20

    return [

    21

        # in/out bandwidth in bytes/s

    22

        'DEF:in_%(idx)d=%(rrd)s:%(ds_in)s:%(cf)s' % params,

    23

        'DEF:out_%(idx)d=%(rrd)s:%(ds_out)s:%(cf)s' % params,

    24

    25

        # total traffic in bits/s

    26

        'CDEF:traf_%(idx)d=in_%(idx)d,out_%(idx)d,+,8,*' % params,

    27

    28

        # average + maximum values

    29

        'VDEF:avg_%(idx)d=traf_%(idx)d,AVERAGE' % params,

    30

        'VDEF:max_%(idx)d=traf_%(idx)d,MAXIMUM' % params,

    31

    32

        # output

    33

        'PRINT:avg_%(idx)d:%(idx)d avg %%lf' % params,

    34

        'PRINT:max_%(idx)d:%(idx)d max %%lf' % params,

    35

    ]

    36

    37

def parse_report (rrds, lines) :

    38

    """

    39

        Parse the report output into a [ (name, avg, max) ] list

    40

    """

    41

    42

    # idx values by type

    43

    data = {

    44

        'avg': {},

    45

        'max': {},

    46

        }

    47

    48

    # interpret output

    49

    for line in lines :

    50

        # parse

    51

        idx, type, value = line.split()

    52

    53

        # convert

    54

        idx = int(idx)

    55

        value = float(value)

    56

    57

        # store

    58

        data[type][idx] = value

    59

    60

    # build into name : (values) dict

    61

    return [

    62

        (

    63

            (name, data['avg'][idx], data['max'][idx])

    64

        ) for idx, name in enumerate(rrds)

    65

    ]

    66

    67

def calc_top_hosts (dir, rrds, ds_in, ds_out, period='15m', count=5, cf='AVERAGE') :

    68

    """

    69

        Return the list of top-N rrd's in the given dir, sorted by total in/out bandwidth average over the given period.

    70

    """

    71

    72

    # vdefs for hosts, avg/max_<idx> in bits/s + prints

    73

    defs = [stmt for idx, name in enumerate(rrds) for stmt in host_def(idx, name, dir, ds_in, ds_out, cf)]

    74

    75

    # execute

    76

    _, _, output, _ = rrd.graph(False, *defs, start=('-%s' % period))

    77

    78

    # parse

    79

    data = parse_report(rrds, output)

    80

    81

    # sort for top-N avg

    82

    data.sort(key=operator.itemgetter(1))

    83

    top_avg = data[:count]

    84

    85

    # sort for top-N max

    86

    data.sort(key=operator.itemgetter(2))

    87

    top_max = data[:count]

    88

    89

    # merge hosts (we lose sorting order)

    90

    return set(host for host, avg, max in top_avg + top_max)

    91

    92