|
1
|
|
|
try: |
|
2
|
|
|
import ujson as json |
|
3
|
|
|
except ImportError: |
|
4
|
|
|
import json |
|
5
|
|
|
|
|
6
|
|
|
import logging |
|
7
|
|
|
|
|
8
|
|
|
try: |
|
9
|
|
|
# Test for mypy support (requires Python 3) |
|
10
|
|
|
from typing import List, Text |
|
11
|
|
|
except: |
|
12
|
|
|
pass |
|
13
|
|
|
|
|
14
|
|
|
|
|
15
|
|
|
class Encoder(object): |
|
16
|
|
|
""" |
|
17
|
|
|
An encoder for the Collectd JSON format |
|
18
|
|
|
See https://collectd.org/wiki/index.php/JSON |
|
19
|
|
|
|
|
20
|
|
|
Sample measurements: |
|
21
|
|
|
|
|
22
|
|
|
[{"values":[0],"dstypes":["derive"],"dsnames":["value"],"time":1436372292.412,"interval":10.000,"host":"26f2fc918f50","plugin":"cpu","plugin_instance":"1","type":"cpu","type_instance":"interrupt"}] |
|
23
|
|
|
|
|
24
|
|
|
[ |
|
25
|
|
|
{ |
|
26
|
|
|
"values": [1901474177], |
|
27
|
|
|
"dstypes": ["counter"], |
|
28
|
|
|
"dsnames": ["value"], |
|
29
|
|
|
"time": 1280959128, |
|
30
|
|
|
"interval": 10, |
|
31
|
|
|
"host": "leeloo.octo.it", |
|
32
|
|
|
"plugin": "cpu", |
|
33
|
|
|
"plugin_instance": "0", |
|
34
|
|
|
"type": "cpu", |
|
35
|
|
|
"type_instance": "idle" |
|
36
|
|
|
} |
|
37
|
|
|
] |
|
38
|
|
|
|
|
39
|
|
|
The following measurement format is also supported, which has more than one value for each sample. |
|
40
|
|
|
[{"values":[0.2, 0.3],"dstypes":["derive"],"dsnames":["cpu_usage", "mem_usage"],"time":1436372292.412,"interval":10.000,"host":"26f2fc918f50","plugin":"cpu","plugin_instance":"1","type":"cpu","type_instance":"interrupt"}] |
|
41
|
|
|
""" |
|
42
|
|
|
|
|
43
|
|
|
def encode(self, msg): |
|
44
|
|
|
# type: (bytes) -> List[Text] |
|
45
|
|
|
measurements = [] |
|
46
|
|
|
|
|
47
|
|
|
for line in msg.decode().split("\n"): |
|
48
|
|
|
try: |
|
49
|
|
|
# Set flag for float precision to get the same |
|
50
|
|
|
# results for Python 2 and 3. |
|
51
|
|
|
json_object = self.parse_line(line) |
|
52
|
|
|
except ValueError as e: |
|
53
|
|
|
logging.debug("Error in encoder: %s", e) |
|
54
|
|
|
continue |
|
55
|
|
|
for entry in json_object: |
|
56
|
|
|
try: |
|
57
|
|
|
# to set plugin, plugin_instance as the measurement name, just need pass ['plugin', 'plugin_instance'] |
|
58
|
|
|
measurement = Encoder.format_measurement_name( |
|
59
|
|
|
entry, ['plugin', 'plugin_instance', 'type']) |
|
60
|
|
|
tags = Encoder.format_tags( |
|
61
|
|
|
entry, ['host', 'type_instance']) |
|
62
|
|
|
value = Encoder.format_value(entry) |
|
63
|
|
|
time = Encoder.format_time(entry) |
|
64
|
|
|
measurements.append(Encoder.compose_data( |
|
65
|
|
|
measurement, tags, value, time)) |
|
66
|
|
|
except Exception as e: |
|
67
|
|
|
logging.debug("Error in input data: %s. Skipping.", e) |
|
68
|
|
|
continue |
|
69
|
|
|
return measurements |
|
70
|
|
|
|
|
71
|
|
|
@staticmethod |
|
72
|
|
|
def parse_line(line): |
|
73
|
|
|
# return json.loads(line, {'precise_float': True}) |
|
74
|
|
|
# for influxdb version > 0.9, timestamp is an integer |
|
75
|
|
|
return json.loads(line) |
|
76
|
|
|
|
|
77
|
|
|
# following methods are added to support customizing measurement name, tags much more flexible |
|
78
|
|
|
@staticmethod |
|
79
|
|
|
def compose_data(measurement, tags, value, time): |
|
80
|
|
|
data = "{0!s},{1!s} {2!s} {3!s}".format(measurement, tags, value, time) |
|
81
|
|
|
return data |
|
82
|
|
|
|
|
83
|
|
|
@staticmethod |
|
84
|
|
|
def format_measurement_name(entry, args): |
|
85
|
|
|
name = [] |
|
86
|
|
|
for arg in args: |
|
87
|
|
|
if arg in entry: |
|
88
|
|
|
# avoid to add extra _ if some entry value is None |
|
89
|
|
|
if entry[arg] != '': |
|
90
|
|
|
name.append(entry[arg]) |
|
91
|
|
|
return '_'.join(name) |
|
92
|
|
|
|
|
93
|
|
|
@staticmethod |
|
94
|
|
|
def format_tags(entry, args): |
|
95
|
|
|
tag = [] |
|
96
|
|
|
for arg in args: |
|
97
|
|
|
if arg in entry: |
|
98
|
|
|
# to avoid add None as tag value |
|
99
|
|
|
if entry[arg] != '': |
|
100
|
|
|
tag.append("{0!s}={1!s}".format(arg, entry[arg])) |
|
101
|
|
|
return ','.join(tag) |
|
102
|
|
|
|
|
103
|
|
|
@staticmethod |
|
104
|
|
|
def format_time(entry): |
|
105
|
|
|
return int(float(entry['time'])) |
|
106
|
|
|
|
|
107
|
|
|
@staticmethod |
|
108
|
|
|
def format_value(entry): |
|
109
|
|
|
values = entry['values'] |
|
110
|
|
|
if len(values) == 1: |
|
111
|
|
|
return "value={0!s}".format(entry['values'][0]) |
|
112
|
|
|
else: |
|
113
|
|
|
# influxdb supports writing a record with multiple field values. |
|
114
|
|
|
# e.g: 'cpu_load_short,host=server01,region=us-west mem=0.1,cpu=0.2 1422568543702900257' |
|
115
|
|
|
field_pairs = [] |
|
116
|
|
|
for key, value in zip(entry['dsnames'], values): |
|
117
|
|
|
field_pairs.append("{0!s}={1!s}".format(key, value)) |
|
118
|
|
|
return ','.join(field_pairs) |
|
119
|
|
|
|
mre /
kafka-influxdb
