

API Docs - v1.0.1

Sink

prometheus (Sink)

The sink publishes events processed by Siddhi into Prometheus metrics and exposes them to Prometheus server at the provided url. The created metrics can be published to Prometheus through 'server' or 'pushGateway' publishing modes depending on the preference of the user. The server mode exposes the metrics through an http server at the provided url and the pushGateway mode pushes the metrics to pushGateway which must be running at the provided url.
 The metric types that are supported by Prometheus sink are counter, gauge, histogram and summary. The values and labels of the Prometheus metrics can be updated through the events.

Syntax

@sink(type="prometheus", job="<STRING>", publish.mode="<STRING>", push.url="<STRING>", server.url="<STRING>", metric.type="<STRING>", metric.help="<STRING>", metric.name="<STRING>", buckets="<STRING>", quantiles="<STRING>", quantile.error="<DOUBLE>", value.attribute="<STRING>", push.operation="<STRING>", grouping.key="<STRING>", @map(...)))

QUERY PARAMETERS

Name	Description	Default Value	Possible Data Types	Optional	Dynamic
job	This parameter specifies the job name of the metric. The name must be the same job name as defined in the prometheus configuration file.	siddhiJob	STRING	Yes	No
publish.mode	This parameter specifies the mode of exposing metrics to Prometheus server.The possible publishing modes are 'server' and 'pushgateway'.	server	STRING	Yes	No
push.url	This parameter specifies the target url of the Prometheus pushGateway where the pushGateway must be listening. This url should be previously defined in prometheus configuration file as a target.	http://localhost:9091	STRING	Yes	No
server.url	This parameter specifies the url where the http server is initiated to expose metrics for 'server' publish mode. This url must be previously defined in prometheus configuration file as a target.	http://localhost:9080	STRING	Yes	No
metric.type	The type of Prometheus metric that has to be created at the sink.  The supported metric types are 'counter', 'gauge', 'histogram' and 'summary'.		STRING	No	No
metric.help	A brief description of the metric and its purpose.		STRING	Yes	No
metric.name	This parameter specifies the user preferred name for the metric. The metric name must match the regex format, i.e., [a-zA-Z_:][a-zA-Z0-9_:]*.		STRING	Yes	No
buckets	The bucket values preferred by the user for histogram metrics. The bucket values must be in 'string' format with each bucket value separated by a comma. The expected format of the parameter is as follows: "2,4,6,8"	null	STRING	Yes	No
quantiles	The user preferred quantile values for summary metrics. The quantile values must be in 'string' format with each quantile value separated by a comma. The expected format of the parameter is as follows: "0.5,0.75,0.95"	null	STRING	Yes	No
quantile.error	The error tolerance value for calculating quantiles in summary metrics. This must be a positive value though less than 1.	0.001	DOUBLE	Yes	No
value.attribute	The name of the attribute in stream definition which specifies the metric value. The defined value attribute must be included inside the stream attributes. The value of the 'value' attribute that is published through events, increase the metric value for the counter and gauge metric types. For histogram and summary metric types, the values are observed.	value	STRING	Yes	No
push.operation	This parameter defines the mode for pushing metrics to pushGateway The available push operations are 'push' and 'pushadd'. The operations differ according to the existing metrics in pushGateway where 'push' operation replaces the existing metrics and 'pushadd' operation only updates the newly created metrics.	pushadd	STRING	Yes	No
grouping.key	This parameter specifies the grouping key of created metrics in key-value pairs. Grouping key is used only in pushGateway mode in order to distinguish the metrics from already existing metrics. The expected format of the grouping key is as follows:  "'key1:value1','key2:value2'"		STRING	Yes	No

System Parameters

Name	Description	Default Value	Possible Parameters
jobName	This is the property that specifies the default job name for the metric. The name must be the same job name as defined in the prometheus configuration file.	siddhiJob	Any string
publishMode	The default publish mode for the Prometheus sink for exposing metrics to Prometheus server. The mode can be either 'server' or 'pushgateway'.	server	server or pushgateway
serverURL	This property configures the url where the http server will be initiated to expose metrics. This url must be previously defined in prometheus configuration file as a target to be identified by Prometheus. By default, the http server will be initiated at 'http://localhost:9080'	http://localhost:9080	Any valid URL
pushURL	This property configures the target url of Prometheus pushGateway where the pushGateway must be listening. This url should be previously defined in prometheus configuration file as a target to be identified by Prometheus.	http://localhost:9091	Any valid URL
groupingKey	This property configures the grouping key of created metrics in key-value pairs. Grouping key is used only in pushGateway mode in order to distinguish the metrics from already existing metrics under same job. The expected format of the grouping key is as follows: "'key1:value1','key2:value2'" .	null	Any key value pairs in the supported format

Examples EXAMPLE 1

@sink(type='prometheus',job='fooOrderCount', server.url ='http://localhost:9080', publish.mode='server', metric.type='counter', metric.help= 'Number of foo orders', @map(type='keyvalue'))
define stream FooCountStream (Name String, quantity int, value int);

In the above example, the Prometheus-sink creates a counter metric with the stream name and defined attributes as labels. The metric is exposed through an http server at the target url.

EXAMPLE 2

@sink(type='prometheus',job='inventoryLevel', push.url='http://localhost:9080', publish.mode='pushGateway', metric.type='gauge', metric.help= 'Current level of inventory', @map(type='keyvalue'))
define stream InventoryLevelStream (Name String, value int);

In the above example, the Prometheus-sink creates a gauge metric with the stream name and defined attributes as labels.The metric is pushed to Prometheus pushGateway at the target url.

Source

prometheus (Source)

The source consumes Prometheus metrics which are exported from the specified url as Siddhi events, by making http requests to the url. According to the source configuration, it analyses metrics from the text response and sends them as Siddhi events through key-value mapping.The user can retrieve metrics of types including, counter, gauge, histogram and summary. Since the source retrieves the metrics from a text response of the target, it is advised to use 'string' as the attribute type for the attributes that correspond to Prometheus metric labels. Further, the Prometheus metric value is passed through the event as 'value'. Therefore, it is advisable to have an attribute with the name 'value' in the stream.
The supported types for the attribute, 'value' are INT, LONG, FLOAT and DOUBLE.

Syntax

@source(type="prometheus", target.url="<STRING>", scrape.interval="<INT>", scrape.timeout="<INT>", scheme="<STRING>", metric.name="<STRING>", metric.type="<STRING>", username="<STRING>", password="<STRING>", client.truststore.file="<STRING>", client.truststore.password="<STRING>", headers="<STRING>", job="<STRING>", instance="<STRING>", grouping.key="<STRING>", @map(...)))

QUERY PARAMETERS

Name	Description	Default Value	Possible Data Types	Optional	Dynamic
target.url	This property specifies the target url where the Prometheus metrics are exported in text format.		STRING	No	No
scrape.interval	This property specifies the time interval in seconds within which the source should make an HTTP request to the provided target url.	60	INT	Yes	No
scrape.timeout	This property is the time duration in seconds for a scrape request to get timed-out if the server at the url does not respond.	10	INT	Yes	No
scheme	This property specifies the scheme of the target URL.  The supported schemes are 'HTTP' and 'HTTPS'.	HTTP	STRING	Yes	No
metric.name	This property specifies the name of the metrics that are to be fetched. The metric name must match the regex format, i.e., [a-zA-Z_:][a-zA-Z0-9_:]* .	Stream name	STRING	Yes	No
metric.type	This property specifies the type of the Prometheus metric that is required to be fetched.  The supported metric types are 'counter', 'gauge'," 'histogram' and 'summary'.		STRING	No	No
username	This property specifies the username that has to be added in the authorization header of the HTTP request, if basic authentication is enabled at the target. It is required to specify both username and password to enable basic authentication. If one of the parameter is not given by user then an error is logged in the console.		STRING	Yes	No
password	This property specifies the password that has to be added in the authorization header of the request, if the basic authentication is enabled at the target. It is required to specify both the username and password to enable basic authentication. If one of the parameter is not given by user, then an error is logged in the console.		STRING	Yes	No
client.truststore.file	The file path to the location of the truststore to which the client needs to send https requests through 'https' protocol.		STRING	Yes	No
client.truststore.password	The password for client-truststore to send https requests. A custom password can be specified if required.		STRING	Yes	No
headers	Headers that should be included as HTTP request headers in the request. The format of the supported input is as follows, "'header1:value1','header2:value2'"		STRING	Yes	No
job	This property defines the job name of the exported Prometheus metrics that has to be fetched.		STRING	Yes	No
instance	This property defines the instance of the exported Prometheus metrics that has to be fetched.		STRING	Yes	No
grouping.key	This parameter specifies the grouping key of the required metrics in key-value pairs. Grouping key is used if the metrics are exported by Prometheus pushGateway in order to distinguish the metrics from already existing metrics.  The expected format of the grouping key is as follows: "'key1:value1','key2:value2'"		STRING	Yes	No

System Parameters

Name	Description	Default Value	Possible Parameters
scrapeInterval	The default time interval in seconds for the Prometheus source to make HTTP requests to the target URL.	60	Any integer value
scrapeTimeout	This default time duration (in seconds) for an HTTP request to time-out if the server at the URL does not respond.	10	Any integer value
scheme	The scheme of the target for Prometheus source to make HTTP requests. The supported schemes are HTTP and HTTPS.	HTTP	HTTP or HTTPS
username	The username that has to be added in the authorization header of the HTTP request, if basic authentication is enabled at the target. It is required to specify both username and password to enable basic authentication. If one of the parameter is not given by user then an error is logged in the console.		Any string
password	The password that has to be added in the authorization header of the HTTP request, if basic authentication is enabled at the target. It is required to specify both username and password to enable basic authentication. If one of the parameter is not given by user then an error is logged in the console.		Any string
trustStoreFile	The default file path to the location of truststore that the client needs to send for HTTPS requests through 'HTTPS' protocol.	${carbon.home}/resources/security/client-truststore.jks	Any valid path for the truststore file
trustStorePassword	The default password for the client-truststore to send HTTPS requests.	wso2carbon	Any string
headers	The headers that should be included as HTTP request headers in the scrape request. The format of the supported input is as follows, "'header1:value1','header2:value2'"		Any valid http headers
job	The default job name of the exported Prometheus metrics that has to be fetched.		Any valid job name
instance	The default instance of the exported Prometheus metrics that has to be fetched.		Any valid instance name
groupingKey	The default grouping key of the required Prometheus metrics in key-value pairs. Grouping key is used if the metrics are exported by Prometheus pushGateway in order to distinguish the metrics from already existing metrics. The expected format of the grouping key is as follows: "'key1:value1','key2:value2'"		Any valid grouping key pairs

Examples EXAMPLE 1

@source(type= 'prometheus', target.url= 'http://localhost:9080/metrics', metric.type= 'counter', metric.name= 'sweet_production_counter', @map(type= 'keyvalue'))
define stream FooStream1(metric_name string, metric_type string, help string, subtype string, name string, quantity string, value double);

In this example, the prometheus source makes an http request to the 'target.url' and analyse the response. From the analysed response, the source retrieves the Prometheus counter metrics with the name, 'sweet_production_counter' and converts the filtered metrics into Siddhi events using the key-value mapper.
The generated maps will have keys and values as follows:
  metric_name -> sweet_production_counter
  metric_type -> counter
  help -> <help_string_of_metric>
  subtype -> null
  name -> <value_of_label_name>
  quantity -> <value_of_label_quantity>
  value -> <value_of_metric>

EXAMPLE 2

@source(type= 'prometheus', target.url= 'http://localhost:9080/metrics', metric.type= 'summary', metric.name= 'sweet_production_summary', @map(type= 'keyvalue'))
 define stream FooStream2(metric_name string, metric_type string, help string, subtype string, name string, quantity string, quantile string, value double);

In this example, the prometheus source makes an http request to the 'target.url' and analyses the response. From the analysed response, the source retrieves the Prometheus summary metrics with the name, 'sweet_production_summary' and converts the filtered metrics into Siddhi events using the key-value mapper.
The generated maps have keys and values as follows:
  metric_name -> sweet_production_summary
  metric_type -> summary
  help -> <help_string_of_metric>
  subtype -> <'sum'/'count'/'null'>
  name -> <value_of_label_name>
  quantity -> <value_of_label_quantity>
  quantile -> <value of the quantile>
  value -> <value_of_metric>

EXAMPLE 3

@source(type= 'prometheus', target.url= 'http://localhost:9080/metrics', metric.type= 'histogram', metric.name= 'sweet_production_histogram', @map(type= 'keyvalue'))
define stream FooStream3(metric_name string, metric_type string, help string, subtype string, name string, quantity string, le string, value double);

In this example, the prometheus source will make an http request to the 'target.url' and analyse the response. From the analysed response, the source retrieves the Prometheus histogram metrics with name 'sweet_production_histogram' and converts the filtered metrics into Siddhi events using the key-value mapper.
The generated maps will have keys and values as follows,
  metric_name -> sweet_production_histogram
  metric_type -> histogram
  help -> <help_string_of_metric>
  subtype -> <'sum'/'count'/'bucket'>
  name -> <value_of_label_name>
  quantity -> <value_of_label_quantity>
  le -> <value of the bucket>
  value -> <value_of_metric>