This output lets you aggregate and send metric data to AWS CloudWatch
This plugin is intended to be used on a logstash indexer agent (but that
is not the only way, see below.) In the intended scenario, one cloudwatch
output plugin is configured, on the logstash indexer node, with just AWS API
credentials, and possibly a region and/or a namespace. The output looks
for fields present in events, and when it finds them, it uses them to
calculate aggregate statistics. If the metricname
option is set in this
output, then any events which pass through it will be aggregated & sent to
CloudWatch, but that is not recommended. The intended use is to NOT set the
metricname option here, and instead to add a CW_metricname
field (and other
fields) to only the events you want sent to CloudWatch.
When events pass through this output they are queued for background
aggregation and sending, which happens every minute by default. The
queue has a maximum size, and when it is full aggregated statistics will be
sent to CloudWatch ahead of schedule. Whenever this happens a warning
message is written to logstash’s log. If you see this you should increase
the queue_size
configuration option to avoid the extra API calls. The queue
is emptied every time we send data to CloudWatch.
Note: when logstash is stopped the queue is destroyed before it can be processed. This is a known limitation of logstash and will hopefully be addressed in a future version.
There are two ways to configure this plugin, and they can be used in combination: event fields & per-output defaults
Event Field configuration…
You add fields to your events in inputs & filters and this output reads
those fields to aggregate events. The names of the fields read are
configurable via the field_*
options.
Per-output defaults… You set universal defaults in this output plugin’s configuration, and if an event does not have a field for that option then the default is used.
Notice, the event fields take precedence over the per-output defaults.
At a minimum events must have a “metric name” to be sent to CloudWatch.
This can be achieved either by providing a default here OR by adding a
CW_metricname
field. By default, if no other configuration is provided
besides a metric name, then events will be counted (Unit: Count, Value: 1)
by their metric name (either a default or from their CW_metricname
field)
Other fields which can be added to events to modify the behavior of this
plugin are, CW_namespace
, CW_unit
, CW_value
, and
CW_dimensions
. All of these field names are configurable in
this output. You can also set per-output defaults for any of them.
See below for details.
Read more about AWS CloudWatch, and the specific of API endpoint this output uses, PutMetricData
output {
cloudwatch {
access_key_id => ... # string (optional)
aws_credentials_file => ... # string (optional)
codec => ... # codec (optional), default: "plain"
dimensions => ... # hash (optional)
field_dimensions => ... # string (optional), default: "CW_dimensions"
field_metricname => ... # string (optional), default: "CW_metricname"
field_namespace => ... # string (optional), default: "CW_namespace"
field_unit => ... # string (optional), default: "CW_unit"
field_value => ... # string (optional), default: "CW_value"
metricname => ... # string (optional)
namespace => ... # string (optional), default: "Logstash"
proxy_uri => ... # string (optional)
queue_size => ... # number (optional), default: 10000
region => ... # string, one of ["us-east-1", "us-west-1", "us-west-2", "eu-west-1", "ap-southeast-1", "ap-southeast-2", "ap-northeast-1", "sa-east-1", "us-gov-west-1"] (optional), default: "us-east-1"
secret_access_key => ... # string (optional)
timeframe => ... # string (optional), default: "1m"
unit => ... # string, one of ["Seconds", "Microseconds", "Milliseconds", "Bytes", "Kilobytes", "Megabytes", "Gigabytes", "Terabytes", "Bits", "Kilobits", "Megabits", "Gigabits", "Terabits", "Percent", "Count", "Bytes/Second", "Kilobytes/Second", "Megabytes/Second", "Gigabytes/Second", "Terabytes/Second", "Bits/Second", "Kilobits/Second", "Megabits/Second", "Gigabits/Second", "Terabits/Second", "Count/Second", "None"] (optional), default: "Count"
use_ssl => ... # boolean (optional), default: true
value => ... # string (optional), default: "1"
workers => ... # number (optional), default: 1
}
}
This plugin uses the AWS SDK and supports several ways to get credentials, which will be tried in this order…
1. Static configuration, using access_key_id
and secret_access_key
params in logstash plugin config
2. External credentials file specified by aws_credentials_file
3. Environment variables AWS_ACCESS_KEY_ID
and AWS_SECRET_ACCESS_KEY
4. Environment variables AMAZON_ACCESS_KEY_ID
and AMAZON_SECRET_ACCESS_KEY
5. IAM Instance Profile (available when running inside EC2)
Path to YAML file containing a hash of AWS credentials.
This file will only be loaded if access_key_id
and
secret_access_key
aren’t set. The contents of the
file should look like this:
:access_key_id: "12345"
:secret_access_key: "54321"
The codec used for output data. Output codecs are a convenient method for encoding your data before it leaves the output, without needing a separate filter in your Logstash pipeline.
The default dimensions [ name, value, … ] to use for events which do not have a CW_dimensions
field
Only handle events without any of these tags. Note this check is additional to type and tags.
The name of the field used to set the dimensions on an event metric
The field named here, if present in an event, must have an array of
one or more key & value pairs, for example…
add_field => [ “CW_dimensions”, “Environment”, “CW_dimensions”, “prod” ]
or, equivalently…
add_field => [ “CW_dimensions”, “Environment” ]
add_field => [ “CW_dimensions”, “prod” ]
The name of the field used to set the metric name on an event
The author of this plugin recommends adding this field to events in inputs &
filters rather than using the per-output default setting so that one output
plugin on your logstash indexer can serve all events (which of course had
fields set on your logstash shippers.)
The name of the field used to set a different namespace per event
Note: Only one namespace can be sent to CloudWatch per API call
so setting different namespaces will increase the number of API calls
and those cost money.
The name of the field used to set the unit on an event metric
The name of the field used to set the value (float) on an event metric
The default metric name to use for events which do not have a CW_metricname
field.
Beware: If this is provided then all events which pass through this output will be aggregated and
sent to CloudWatch, so use this carefully. Furthermore, when providing this option, you
will probably want to also restrict events from passing through this output using event
type, tag, and field matching
The default namespace to use for events which do not have a CW_namespace
field
URI to proxy server if required
How many events to queue before forcing a call to the CloudWatch API ahead of timeframe
schedule
Set this to the number of events-per-timeframe you will be sending to CloudWatch to avoid extra API calls
The AWS Region
The AWS Secret Access Key
Only handle events with all of these tags. Note that if you specify a type, the event must also match that type. Optional.
Constants
aggregate_key members
Units
How often to send data to CloudWatch
This does not affect the event timestamps, events will always have their
actual timestamp (to-the-minute) sent to CloudWatch.
We only call the API if there is data to send.
See the Rufus Scheduler docs for an explanation of allowed values
The type to act on. If a type is given, then this output will only act on messages with the same type. See any input plugin’s “type” attribute for more. Optional.
The default unit to use for events which do not have a CW_unit
field
If you set this option you should probably set the “value” option along with it
Should we require (true) or disable (false) using SSL for communicating with the AWS API
The AWS SDK for Ruby defaults to SSL so we preserve that
The default value to use for events which do not have a CW_value
field
If provided, this must be a string which can be converted to a float, for example…
“1”, “2.34”, “.5”, and “0.67”
If you set this option you should probably set the unit
option along with it
The number of workers to use for this output. Note that this setting may not be useful for all outputs.