Count Telemetry

Counts the logs, metric data points, or trace spans matching an OTTL condition over a fixed interval and emits the result as a new metric. The created metric's name and units are configurable, and fields from the matched telemetry can be carried onto the metric as attributes. A common use is converting high-volume logs into a low-cardinality count metric so the original logs can be dropped while the signal is retained.

Not available in v2 configurations

This processor cannot be added to a v2 configuration, which is the default for all new configurations. Use the Count connector instead. Existing v1 configurations keep working.

Supported Telemetry Types

Metrics	Logs	Traces
✓	✓	✓

A single instance can count more than one signal at once. Choose Telemetry Type selects which signals are active, and each selected signal is configured independently below.

Configuration

General

Parameter	Type	Required	Default	Description
Choose Telemetry Type	Telemetry Selector	No	Logs, Metrics, Traces	Which signals this instance counts. Each selected signal exposes its own match expression, metric name, units, and attribute settings.

Logs

Shown when Choose Telemetry Type includes Logs.

Parameter	Type	Required	Default	Description
Match Expression	OTTL Condition	No	(empty)	OTTL expression to find matching logs. Uses the log context. Empty counts all logs.
Metric Name	String	Yes	log.count	Name to give the created metric.
Metric Units	String	Yes	{logs}	Units to give the created metric. See UCUM for available units.
Enable Attributes	Boolean	No	false	Enable to add attributes to the created metric.
Metric Attributes	Map	Yes *	(empty)	Attributes built from OTTL expressions, keyed by attribute name. Shown when Enable Attributes is on.

* Required only when Enable Attributes is on.

Metrics

Shown when Choose Telemetry Type includes Metrics.

Parameter	Type	Required	Default	Description
Match Expression	OTTL Condition	No	(empty)	OTTL expression to find matching metric data points. Uses the datapoint context. Empty counts all data points.
Metric Name	String	Yes	datapoint.count	Name to give the created metric.
Metric Units	String	Yes	{datapoints}	Units to give the created metric. See UCUM for available units.
Enable Attributes	Boolean	No	false	Enable to add attributes to the created metric.
Metric Attributes	Map	Yes *	(empty)	Attributes built from OTTL expressions, keyed by attribute name. Shown when Enable Attributes is on.

* Required only when Enable Attributes is on.

Traces

Shown when Choose Telemetry Type includes Traces.

Parameter	Type	Required	Default	Description
Match Expression	OTTL Condition	No	(empty)	OTTL expression to find matching trace spans. Uses the span context. Empty counts all spans.
Metric Name	String	Yes	span.count	Name to give the created metric.
Metric Units	String	Yes	{spans}	Units to give the created metric. See UCUM for available units.
Enable Attributes	Boolean	No	false	Enable to add attributes to the created metric.
Metric Attributes	Map	Yes *	(empty)	Attributes built from OTTL expressions, keyed by attribute name. Shown when Enable Attributes is on.

* Required only when Enable Attributes is on.

Advanced

Parameter	Type	Required	Default	Description
Interval	Integer	No	60	Interval, in seconds, to count telemetry over. Each interval emits one data point per unique attribute set.

Examples

Count HTTP requests by status code (logs)

Counts HTTP server access logs and breaks the count out by status code, so 500s stay visible after the underlying logs are dropped. The logs are JSON with a status field in the body.

The match expression excludes any log without a status code, and the status code is carried onto the count metric as an attribute named status_code:

apiVersion: bindplane.observiq.com/v1
kind: Processor
metadata:
  name: count-http-by-status
spec:
  type: count_telemetry
  parameters:
    - name: telemetry_types
      value:
        - Logs
    - name: log_match
      value: body["status"] != nil
    - name: log_metric_name
      value: http.request.count
    - name: log_metric_units
      value: "{requests}"
    - name: log_enable_attributes
      value: true
    - name: log_attributes
      value:
        status_code: body["status"]
        method: body["method"]

Count error spans by service (traces)

Counts only spans with an error status, broken out by service name, emitting a metric every 30 seconds:

apiVersion: bindplane.observiq.com/v1
kind: Processor
metadata:
  name: count-error-spans
spec:
  type: count_telemetry
  parameters:
    - name: telemetry_types
      value:
        - Traces
    - name: span_match
      value: status.code == STATUS_CODE_ERROR
    - name: span_metric_name
      value: span.error.count
    - name: span_metric_units
      value: "{spans}"
    - name: span_enable_attributes
      value: true
    - name: span_attributes
      value:
        service: resource.attributes["service.name"]
    - name: interval
      value: 30

Count data points above a threshold (metrics)

Counts metric data points whose value exceeds a threshold, useful for tracking how often a gauge breaches a limit:

apiVersion: bindplane.observiq.com/v1
kind: Processor
metadata:
  name: count-high-datapoints
spec:
  type: count_telemetry
  parameters:
    - name: telemetry_types
      value:
        - Metrics
    - name: datapoint_match
      value: value_double > 90.0
    - name: datapoint_metric_name
      value: datapoint.threshold.count
    - name: datapoint_metric_units
      value: "{datapoints}"

Configuration Tips

• Leaving a Match Expression empty counts every record of that signal. Add a condition to count only the subset you care about.
• The Interval (under Advanced) controls how often a count is emitted and the window each count accumulates over. A longer interval emits fewer, coarser data points and holds the counted series in memory until the window closes; a shorter interval emits more frequently at finer resolution but puts more data points on the metrics pipeline. Pick it against the time resolution you need versus the metric volume the pipeline can tolerate.
• Each unique combination of metric attribute values produces its own counted series. Building attributes from high-cardinality fields (request IDs, raw URLs, user IDs) multiplies the emitted series and can drive up metric storage cost. Prefer low-cardinality keys like status code, method, or service name.
• The created count metric appears on the metrics pipeline. When counting logs or traces, make sure your configuration routes those new metrics to a metrics destination.
• This processor converts telemetry into counts but does not drop the original telemetry. Pair it with a filter or sampling processor downstream if the goal is to reduce volume.

Troubleshooting

No count metric is produced

Symptoms: the expected count metric never appears at the destination.

Solutions:

1. Confirm the signal is selected in Choose Telemetry Type and that records of that signal actually flow through the pipeline.
2. Verify the Match Expression evaluates true for at least some records. An expression that never matches yields no count.
3. Confirm the metrics output is routed to a metrics destination, especially when counting logs or traces.

The count is lower or higher than expected

Symptoms: the emitted count does not match the observed telemetry volume.

Solutions:

1. Check the Match Expression. A condition that is too narrow undercounts; an empty expression counts everything.
2. Remember each data point covers one Interval (default 60 seconds). Compare counts over matching time windows.

Counts arrive late or only at interval boundaries

Symptoms: count metrics appear in steps at interval boundaries rather than continuously.

Solution: this is inherent to interval aggregation. A count is emitted only when its Interval closes, so a 60-second interval can delay a data point by up to 60 seconds. Lower the Interval to emit more frequently at finer resolution, accepting more data points on the metrics pipeline.

Too many metric series

Symptoms: the count metric explodes into many series and increases cost.

Solutions:

1. Review the Metric Attributes expressions and remove high-cardinality fields.
2. Disable Enable Attributes to emit a single aggregate count per signal.

Standalone Processor

apiVersion: bindplane.observiq.com/v1
kind: Processor
metadata:
  name: count-telemetry
spec:
  type: count_telemetry
  parameters:
    - name: telemetry_types
      value:
        - Logs
        - Metrics
        - Traces
    - name: log_match
      value: body["status"] != nil
    - name: log_metric_name
      value: log.count
    - name: log_metric_units
      value: "{logs}"
    - name: log_enable_attributes
      value: true
    - name: log_attributes
      value:
        status_code: body["status"]
    - name: datapoint_match
      value: ""
    - name: datapoint_metric_name
      value: datapoint.count
    - name: datapoint_metric_units
      value: "{datapoints}"
    - name: span_match
      value: status.code == STATUS_CODE_ERROR
    - name: span_metric_name
      value: span.count
    - name: span_metric_units
      value: "{spans}"
    - name: interval
      value: 60

Related Resources

• Count Connector — OpenTelemetry Collector Contrib
• OTTL — OpenTelemetry Transformation Language

Bindplane Resources

• Filter by Condition