EUM Server and Metrics Configuration · inspectIT Ocelot Documentation

The configuration file defines the mapping between the concrete Boomerang metric and a OpenCensus metric, as the following sample configuration file shows:

inspectit-eum-server:
  definitions:
     page_ready_time:
      measure-type: LONG
      value-expression: "{t_page}"
      unit: ms
      views:
        '[page_ready_time/SUM]': {aggregation: SUM}
        '[page_ready_time/COUNT]': {aggregation: COUNT}

    load_time:
      measure-type: LONG
      value-expression: "{t_done}"
      beacon-requirements:
        - field: rt.quit
          requirement: NOT_EXISTS
      unit: ms
      views:
        '[load_time/SUM]': {aggregation: SUM}
        '[load_time/COUNT]': {aggregation: COUNT}

    calc_load_time:
      measure-type: LONG
      value-expression: "{rt.end} - {rt.tstart}"
      beacon-requirements:
        - field: rt.quit
          requirement: NOT_EXISTS
      unit: ms
      views:
        '[calc_load_time/SUM]': {aggregation: SUM}
        '[calc_load_time/COUNT]': {aggregation: COUNT}

    start_timestamp:
      measure-type: LONG
      value-expression: "{rt.tstart}"
      unit: ms

    navigation_start_timestamp:
      measure-type: LONG
      value-expression: "{rt.nstart}"
      unit: ms

    end_timestamp:
      measure-type: LONG
      value-expression: "{rt.end}"
      unit: ms
      views:
        end_timestamp:
          aggregation: LAST_VALUE
          tags: {APPLICATION: true}

  tags:
    extra:
      APPLICATION: my-application
    
    beacon:
      URL: u
      OS: ua.plt
    
    define-as-global:
      - URL
      - OS
      - COUNTRY_CODE
    
    custom-ip-mapping:
      department-1:
        - 10.10.0.0/16
        - 11.11.0.3
      department-2:
        - 14.14.0.0/16
        - 14.15.0.1
  exporters:
    metrics:
      prometheus:
        enabled: true
        host: localhost
        port: 8888
    tracing:
      jaeger:
        enabled: true
        grpc: localhost:14250
        service-name: browser-js

Metrics Definition

A metric is defined the same way as in the inspectIT Ocelot Java agent. Please see the section Metrics / Custom Metrics for detailed information.

In contrast to the agent's metric definition, the EUM server's metric definition contains additional fields. These additional fields are the following:

Attributes	Note
`value-expression`	An expression used to calculate the measure's value from a beacon.
`beacon-requirements`	Requirements which have to be fulfilled by Beacons. Beacons which do not match all requirements will be ignored by this metric definition.

Smoothed Average View

note

The Smoothed Average View is currently only available in the EUM server.

In addition to the Quantile View, which is already described in the inspectIT Ocelot Java agent configuration, the EUM Server provides a Smoothed Average View. In contrast to the quantiles, it is possible to drop values of a certain time window. This can be useful to deliberately remove outliers before averaging.

note

Please read the section on Quantile Views to get an insight into the data collection. The Smoothed Average View is based on the same principle.

The actual metric configuration are extended in the EUM server by the following properties:

Config Property	Default	Description
`aggregation`	`LAST_VALUE`	Specifies how the measurement data is aggregated in this view. Possible values are `LAST_VALUE`, `COUNT`, `SUM`, `HISTOGRAM` `QUANTILES` and `SMOOTHED_AVERAGE`. Except for `QUANTILES` and `SMOOTHED_AVERAGE`, these correspond to the OpenCensus Aggregations.
`drop-upper`	`0.0`	Required if aggregation is `SMOOTHED_AVERAGE`. Specifies the percentage of the highest values to be dropped before calculating the average.
`drop-lower`	`0.0`	Required if aggregation is `SMOOTHED_AVERAGE`. Specifies the percentage of the lowest values to be dropped before calculating the average.
`time-window`	`${inspectit.metrics.frequency}`	Required if aggregation is `QUANTILES` or `SMOOTHED_AVERAGE`. The time window over which the quantiles or the smoothed average are captured.
`max-buffered-points`	`16384`	Required if aggregation is `QUANTILES` or `SMOOTHED_AVERAGE`. A safety limit defining the maximum number of points to be buffered.

As an example, the following snippet defines a metric with the name load_time and a view loadtime/smoothed: The configuration has the effect of ordering the values in a 1-minute time window by size and dropping the upper 10 percent before calculating the average.

inspectit:
  metrics:
    definitions:
      load_time:
        measure-type: LONG
        value-expression: "{t_done}"
        beacon-requirements:
          - field: rt.quit
            requirement: NOT_EXISTS
        unit: ms
        views:
          '[load_time/smoothed]':
              aggregation: SMOOTHED_AVERAGE
              time-window: 1m
              drop-upper: 0.1
              drop-lower: 0.0

Value Expressions

The value-expression field can be used to specify a field which value is used for the specified metrics. In order to reference a field, the following pattern is used: {FIELD_KEY}. For example, a valid expression, used to extract the value of a field t_load, would be {t_load}.

note

Note that a beacon has to contain all fields referenced by the expression in order to be evaluated and recorded.

Value expressions also support operations for basic arithmetic operations. Thus, to calculate a difference of two beacon fields, the following expression can be used:

  ...
    my-metric:
      ...
      value-expression: "{field.a} - {field.b}"
      ...

Value expression are supporting the following operations:

addition
subtraction
multiplication
division
unary plus/minus
parentheses

Using the operations above, complex calculations can be done, for example:

  ...
    my-metric:
      ...
      value-expression: "- {field.c} * ({field.a} - {field.b}) / {field.a}"
      ...

Beacon Requirements

The beacon-requirements field can be used to specify requirements which have to be fulfilled by the beacons in order to be evaluated by a certain metric. If any requirement does not fit a beacon, the beacon is ignored by the metric.

The following requirements are available:

Type	Note
`EXISTS`	The targeted field must exist.
`NOT_EXISTS`	The targeted field must not exist.
`HAS_INITIATOR`	The beacon must have one of the specified initiators.

Firstly, you can specify that the metric expects a field to exist or not exist in the beacon:

  ...
    my-metric:
      ...
      beacon-requirements:
        - field: rt.quit
          requirement: NOT_EXISTS

In this example, my-metric will only be recorded if the field rt.quit does not exist in the beacon. Alternatively you can use the requirement EXISTS to make sure the metric is only recorded if the given field is present.

Additionally you can specify that you only want to record the metric if the received beacon has a specific initiator:

  ...
    my-metric:
      ...
      beacon-requirements:
        - initiators: [SPA_SOFT, SPA_HARD]
          requirement: HAS_INITIATOR

The available initiators are DOCUMENT for the initial pageload beacons, XHR for Ajax-Beacons and SPA_SOFT and SPA_HARD for soft and hard SPA navigation beacons. The metric will only be recorded for beacons whose http.initiator field matches any of the elements provided in the initiators list.

Additional Beacon Fields

T_OTHER.*

The t_other.* fields are a special set of fields which are resolved based on the content of the beacon's t_other field.

The Boomerang agent allows to set custom-timer data, which represents a arbitrary key-value pair. The key represents the name of the timer and the value the timer's value which may be a duration or any number. This can be used by applications to measure custom durations or events. See the Boomerang's documentation for more information.

When using custom timers, Boomerang combines their values as a comma-separated list and sends them in the t_other attribute. For example, a beacon can be structured as follows: t_other=t_domloaded|437,boomerang|420,boomr_fb|252

In order to use the custom-timer data for metrics or tags, the EUM server will split the timer data contained in the t_other field into separate fields. The syntax of the resulting fields is: t_other.[CUSTOM_TIMTER_NAME]. These resulting fields can be used, for example, as a value input for metrics.

Example

A t_other field containing t_domloaded|437,boomr_fb|252 will produce the following results:

t_other.t_domloaded = 437
t_other.boomr_fb = 252

CLIENT.HEADER.*

The client.header.* fields are resolved based on the headers of the beacons request. Each header will be available at a new attribute with the client.header prefix in the beacon. Note: the capitalisation of the header name is preserved!

Example

Assuming the request for sending a beacon to the EUM server contains the header Accept-Encoding: gzip,deflate and Connection: keep-alive. In this case, the following beacon properties will be generated and accessible via the configuration:

client.header.Accept-Encoding with value gzip,deflate
client.header.Connection with value keep-alive

RT.BMR.*

The rt.bmr.* fields are resolved based on the content of the beacon's rt.bmr field. See the Boomerang's documentation for more information.

The rt.bmr parameter consists of a comma-separated list with the following resource timing information for Boomerang itself:

startTime
responseEnd
responseStart
requestStart
connectEnd
secureConnectionStart
connectStart
domainLookupEnd
domainLookupStart
redirectEnd
redirectStart

Example

A rt.bmr field containing 123,477,,1 will produce the following results:

rt.bmr.startTime = 123
rt.bmr.responseEnd = 477
rt.bmr.responseStart = 0
rt.bmr.requestStart = 1

rt.bmr.connectEnd = 0 *
rt.bmr.secureConnectionStart = 0 *
...

* All values which does not exist in the parameter will be populated with zero!

Tags Definition

We distinguish between to different types of tags:

Attributes	Note
`extra`	Extra tags define tags, which are manually set in the configuration and can be considered as constants.
`beacon`	Beacon tags define tags, whose value is resolved by an incoming beacon entry.

For example, the following configuration specifies the two tags APP and URL. The tag APP will always be resolved to the value my-application, where the tag URL will be resolved to the value of the field u of a received beacon.

inspectit-eum-server:
  tags:
    extra:
      APP: my-application
    beacon:
      URL: 
        input: u

Tags configured via beacon offer some additional flexibility: In addition to simply copying the input value, it is possible to perform one or multiple regular expression replacements.

Example: in case the u attribute contains a URL which is: http://server/user/100. The following configuration can be used to extract the HTTP-Path from it.

inspectit-eum-server:
  tags:
    beacon:
      MY_PATH: 
        input: u
        replacements:
         -  pattern:  '^.*\/\/([^\/]*)([^?]*).*$'
            replacement: "$2"
            keep-no-match: false

The replacements property defines a list of regular expressions and corresponding replacements to apply. They will be applied in the order they are listed. For each list element, the pattern property defines the regex to use for the replacement. All matches of the pattern in the input value are replaced with the string defined by replacement. The keep-no-match option of each entry defines what to do if the given input does not match the given regex at any place. If it is set to true, the previous value is kept unchanged. If it is set to false, the given tag won't be created in case no match is found. Note that capture groups are supported and can be referenced in the replacement string using $1, $2, etc. as shown in the example.

The following example extends the previous one by additionally replacing all user-IDs within the path:

inspectit-eum-server:
  tags:
    beacon:
      MY_PATH: 
        input: u
        replacements:
         -  pattern:  '^.*\/\/([^\/]*)([^?]*).*$'
            replacement: "$2"
            keep-no-match: false
         -  pattern:  '\/user\/\d+'
            replacement: '/user/{id}'

With these settings, the tag will be extracted from u just like in the previous example. However, an additional replacement will be applied afterwards causing user-IDs to be erased from the path. Note that we did not specify keep-no-match for the second replacement. keep-no-match default to true, meaning that the path will be preserved without any changes in case it does not contain any user-IDs.

Using this mechanism, the EUM server provides the following tags out of the box:

Tag	Description
`U_NO_QUERY`	The Boomerang u property but without query parameters. Check out Boomerang.
`U_HOST`	The host specified in the u property. Check out Boomerang.
`U_PORT`	The port specified in the u property. Check out Boomerang.
`U_PATH`	The http path specified in the u property. Check out Boomerang.
`PGU_NO_QUERY`	The Boomerang pgu property but without query parameters. Check out Boomerang.
`PGU_HOST`	The host specified in the pgu property. Check out Boomerang.
`PGU_PORT`	The port specified in the pgu property. Check out Boomerang.
`PGU_PATH`	The http path specified in the pgu property. Check out Boomerang.

Additional Tags

The EUM server provides a set of additional tags which can be used like all other tag. See the following sections for a detailed description of the existing additional tags.

COUNTRY_CODE

The COUNTRY_CODE tag contains the geolocation of the beacon's origin. It is resolved by using the client IP and the GeoLite2 database. If the IP cannot be resolved, the tag value will be empty.

Custom COUNTRY_CODE Mapping

Besides using the internal GeoLite2 database, it is possible to define custom IP mappings. The property custom-ip-mapping holds a map of possible tag values, which are mapped to certain IPs or CIDRs. The tag values are published with the tag COUNTRY_CODE and have a higher priority than the results of the GeoLite2 database. If the IP cannot be resolved with the custom mapping, the mapping of the GeoLite2 database will be used.

inspectit-eum-server:
  tags:
    custom-ip-mapping:
      department-1:
        - 10.10.0.0/16
        - 11.11.0.3
      department-2:
        - 14.14.0.0/16
        - 14.15.0.1

Global Tags

Tags will not be attached to metrics unless a metric explicitly defines to use a certain tag.

In order to simplify the configuration, it is possible to define tags which are always attached to metrics, even a metric does not explicitly specifies it. This can be achieved by adding a tag's name to the define-as-global property. Each tag which is listed under this property will be added to each registered metric.

For example, the following configuration causes that each metric will be enriched by a tag called COUNTRY_CODE.

inspectit-eum-server:
  tags:
    define-as-global:
      - COUNTRY_CODE

Resource Timings

important

The resource timing processing is enabled by default and can be disabled by setting the property inspectit-eum-server.resource-timing.enabled to false.

The EUM server can extract information about the resources timings which are reported as part of the Boomerang beacon field restiming. The resource timing information is decompressed from the beacon and exposed as part of the resource_time metric. This metric contains following tags:

Tag	Description
`initiatorType`	The type of the element initiating the loading of the resource. See all initiator types.
`crossOrigin`	If a resource loading is considered as cross-origin request. See more information about CORS.
`cached`	If a resource was cached and loaded from the browser disk or memory storage. Note that cached tag will only be set for same-origin requests, as some resource timing metrics are restricted and will not be provided cross-origin unless the Timing-Allow-Origin header permits.

note

Please note that all global tags will be attached as well. Attaching custom tags works in the same way as defining metrics.

For example, the following configuration causes that the resource timing processing is enabled and will be enriched by a tag called U_HOST.

inspectit-eum-server:
  resource-timing:
    enabled: true
    tags: 
      U_HOST: true

Exporters

Metrics

The EUM server comes with the same Prometheus and InfluxDB exporter as the Ocelot agent. The exporter's configurations options are the same as for the agent. However, they are located under the inspectit-eum-server.exporters.metrics configuration path.

Prometheus

By default, the prometheus exporter is enabled and available on port 8888.

The following configuration snippet shows the default configuration of the prometheus-exporter:

inspectit-eum-server:
  exporters:
    metrics:
      prometheus:
        # Determines whether the prometheus exporter is enabled.
        enabled: true

        # The host of the prometheus HTTP endpoint.
        host: localhost

        # The port of the prometheus HTTP endpoint.
        port: 8888

InfluxDB

By default, the InfluxDB exporter is enabled, but it is not active since the url-property is not set. The property can be set via inspectit-eum-server.exporters.metrics.influx.url.

The following configuration snippet makes the InfluxDB Exporter send every 15 seconds metrics to an InfluxDB available under localhost:8086 to the database inspectit:

inspectit-eum-server:
  exporters:
    metrics:
      influx:
        # Determines whether the InfluxDB exporter is enabled.
        enabled: true

        # the export interval of the metrics.
        export-interval: 15s

        # The http url of influx.
        # If this property is not set, the InfluxDB exporter will not be started.
        url: "http://localhost:8086"

        # The database to write to.
        # If this property is not set, the InfluxDB exporter will not be started.
        database: "inspectit"

        # The username to be used to connect to the InfluxDB.
        # username:

        # The password to be used to connect to the InfluxDB.
        # password:

        # The retention policy to write to.
        # If this property is not set, the InfluxDB exporter will not be started.
        retention-policy: "autogen"

        # If true, the specified database will be created with the autogen retention policy.
        create-database: true

        # If disabled, the raw values of each counter will be written to the InfluxDB on each export.
        # When enabled, only the change of the counter in comparison to the previous export will be written.
        # This difference will only be written if the counter has changed (=the difference is non-zero).
        # This can greatly reduce the total data written to InfluxDB and makes writing queries easier.
        counters-as-differences: true

        # The size of the buffer for failed batches.
        # E.g. if the exportInterval is 15s and the buffer-size is 4, the export will keep up to one minute of data in memory.
        buffer-size: 40

Tracing

note

In case Boomerang is used as EUM agent, Note that if you use Boomerang as your EUM agent, it will not capture traces by default. To capture traces with Boomerang, a special tracing plugin must be used.

More information can be found in the chapter on installing the EUM agent.

The EUM server supports trace data forwarding to the Jaeger exporter. The exporter is using the Jaeger Protobuf via gRPC API in order to forward trace data. By default, the Jaeger exporter is enabled, but it is not active since the grpc property is not set.

The following configuration snippet makes the Jaeger exporter send traces to a Jaeger instance avialable under localhost:14250.

inspectit-eum-server:
  exporters:
    tracing:
      jaeger:
      # If jaeger exporter for the OT received spans is enabled.
      enabled: true

      # Location of the jaeger gRPC API.
      # Either a valid NameResolver-compliant URI, or an authority string.
      # If this property is not set, the jaeger-exporter will not be started.
      grpc: localhost:14250

      # service name for all exported spans.
      service-name: browser-js

note

The GRPC property needs to be set without protocol (e.g. localhost:14250)!

Additional Span Attributes

The EUM server is able to enrich a received span with additional attributes. Currently, the following attributes are added to each span.

Attribute	Description
`client.ip`	The sender IP address of the request that sent the spans. This address can be anonymised if required by data protection rules. See Masking Client IP Addresses.

Masking Client IP Addresses

The attribute client.ip is added to each span, which contains the sender IP address of the request with which the span was sent. This IP address can be anonymised with the following configuration:

inspectit-eum-server:
  exporters:
    tracing:
      # Specifies whether client IP addresses which are added to spans should be masked.
      mask-span-ip-addresses: true

If the mask-span-ip-addresses setting is set to true, the last 8 bits are set to 0 for IPv4 addresses. For IPv6 addresses, the last 48 bits are set to 0.

Beacons

The EUM Server supports that received beacons can be exported or sent to other systems. Currently only export via HTTP is supported. In this case, the beacons are sent in JSON format. This allows the received beacons to be sent to an HTTP endpoint (e.g. Logstash).

The following configuration snippet can be used in the EUM server for enabling beacon exportation via HTTP.

inspectit-eum-server:
  exporters:
    beacons:
      http:
        # Whether beacons should be exported via HTTP
        enabled: true
        # The endpoint to which the beacons are to be sent
        endpoint-url: https://localhost:8080
        # The max. amount of threads exporting beacons (min. 1)
        worker-threads: 2
        # The maximum number of beacons to be exported using a single HTTP request (min. 1)
        max-batch-size: 100
        # The flush interval to export beacons in case the 'max-batch-size' has not been reached (min. 1 second)
        flush-interval: 5s
        # When specified, the request will be using this username for Basic authentication
        username: user
        # The password used for Basic authentication
        password: 123

The EUM server uses Basic Authentication for the request if a username is specified. Otherwise no authentication is used.

Self-Monitoring

important

Self-Monitoring is enabled by default and can be disabled by setting the property inspectit-eum-server.self-monitoring.enabled to false.

For the purpose of self-monitoring, the EUM server offers a set of metrics that reflect its state. These metrics are exposed using its Prometheus endpoint which also is used for the EUM beacon data. Currently, the following self monitoring metrics are available.

Metric name	Description
`inspectit_eum_self_beacons_received_count`	Counts the number of received beacons
`inspectit_eum_self_beacons_export_count`	Counts the number of beacon exports
`inspectit_eum_self_beacons_export_duration_sum`	The total duration needed for beacon exports
`inspectit_eum_self_beacons_export_batch_sum`	The number of exported beacons per export
`inspectit_eum_self_beacons_processor_count`	Counts the number of processed beacons
`inspectit_eum_self_beacons_processor_duration_sum`	The total duration needed for beacon processing