Configuration
The configuration format used by ingraind
is TOML.
The primary goal of the configuration system is to set up the grains and the
processing pipelines used by ingraind
. As a rule of thumb, all settings that
are sensitive (eg. credentials), or can change on every deployment, will be
taken from environment variables.
The TOML file has two sections: the probes, and the pipelines.
The probe
array contains the configuration for the probes, and a single entry
looks like this:
[[probe]]
pipelines = ["my_pipeline", "another_pipeline"]
[probe.config]
type = "Files"
monitor_dirs = ["/"]
The probe
has to specify a list of pipelines that will receive its data.
The probe.config
section configures the probe itself; the type
attribute is
mandatory, anything else is probe dependent.
The different brackets in [[probe]]
and [probe.config]
are intentional:
[[probe]]
means a new entry in the probe
array, while [probe.config]
means
"the config
attribute of the probe".
The pipelines are named sequences of actions. Some steps in the pipeline will have state, so be aware that multiple probes targeting the pipeline with the same name, will use the same instance of the pipeline.
Configuration of pipelines looks like this:
[pipeline.my_pipeline.config]
backend = "StatsD"
use_tags = true
[[pipeline.statsd.steps]]
type = "Whitelist"
allow = ["k1", "k2"]
[[pipeline.statsd.steps]]
type = "AddSystemDetails"
[[pipeline.statsd.steps]]
type = "Buffer"
interval_s = 30
The distinction between the [pipeline.statsd.config]
and
[[pipeline.statsd.steps]]
are the same as previously.
Events generated by the probe
will hit the first step in the pipeline. Steps
are defined in sequence, and events will trickle through every step, unless they
are filtered out.
The pipeline.name.config.backend
specifies which backend is used to
transmit or store the events after every step in the pipeline is complete, and
is a mandatory attribute.
Other attributes in the pipeline.name.config
section are dependent on the
given backend.