statsdproxy 0.4.1

A proxy for transforming, pre-aggregating and routing statsd metrics
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62

# Define middlewares to transform statsd metrics before they are forwarded.
# Middlewares are applied top-to-bottom, and middlewares of the same `type` can
# be applied multiple times.
#
# `middlewares: []` will make statsdproxy forward UDP packets almost verbatim,
# regardless of whether the metrics are parseable or not. It's equivalent to
# specifying an empty configuration file or none at all.

middlewares:
  # Remove a list of tag names ("a", "b" and "c") from incoming metrics
  # Also removes tags that start or end with certain words ("foo" or "bar")
  - type: deny-tag
    tags: [a, b, c]
    starts_with: [foo]
    ends_with: [bar]

  # Allow a list of tag names ("a", "b" and "c") from incoming metrics, and
  # remove all other tags.
  - type: allow-tag
    tags: [x, y, z]

  # Apply a limit on the number of timeseries that can be passed through.
  # Multiple limits with different windows can be specified.
  - type: cardinality-limit
    limits:
      - window: 3600
        limit: 3

  # Fold many metrics into one. Currently only gauges and counters are
  # supported, other types or otherwise unparseable lines will be passed
  # through unbuffered.
  - type: aggregate-metrics
    # Whether counters should be aggregated.
    # Defaults to true.
    #
    # aggregate_counters: true

    # Whether gauges should be aggregated.
    # Defaults to true.
    #
    # aggregate_gauges: true

    # Flush the aggregate buffer every `flush_interval` milliseconds.
    # Defaults to 1 second.
    #
    # flush_interval: 1000

    # Normally the times at which metrics are flushed are approximately aligned
    # with a multiple of `flush_interval`. For example, a `flush_interval` of 1
    # hour means that metrics are flushed at the start of every (wall clock)
    # hour. You can change this parameter to shift this bucketing window by a
    # number of seconds, possibly to remove certain artifacts when aggregating
    # at multiple levels. The amount can be negative.
    # Defaults to 0.
    #
    # flush_offset: 0

    # The maximum number of metrics to buffer up. If that limit is hit, the map is forcibly
    # flushed before the flush_interval.
    # Defaults to no limit.
    #
    # max_map_size: ~