Metric extraction stage in OpenPipeline

• Latest Dynatrace
• Explanation
• 5-min read
• Updated on Aug 14, 2026

This article describes the Metric extraction stage in OpenPipeline and the available processors. In the Metric extraction stage, you can extract metrics from incoming records that match a condition.

Processors

The processors in the stage are:

Each processor creates a new metric data point for each matching record. The source record is not modified.

Counter metric

The Counter metric processor increments a counter by 1 for each matching record. Use it to count discrete events or occurrences, such as error responses, user logins, or deployments. The counter always increments by 1; no field extraction is required.

Parameters

Example: Count HTTP errors per status code

The following example counts HTTP error responses and groups them by status code.

The processor applies to all records that match the following condition:

isNotNull(http.status_code) AND toLong(http.status_code) >= 400

The processor is configured as follows:

Unprocessed

{
  "timestamp": "2026-06-01T10:00:00Z",
  "http.method": "GET",
  "http.status_code": "404",
  "service.name": "checkout"
}

The processor creates a counter metric data point log.http.errors{status_code="404"} 1.

Histogram metric

The Histogram metric processor reads a numeric value from a record field and records it as a histogram data point. Use it to track value distributions such as request durations, payload sizes, or queue depths.

To learn more, see Measure an important code path with histograms in OpenPipeline.

Parameters

Example: Track request duration distribution

The following example records the distribution of request durations from log records.

The processor applies to all records that match the following condition:

isNotNull(request.duration_ms)

The processor is configured as follows:

Unprocessed

{
  "timestamp": "2026-06-01T10:00:00Z",
  "service.name": "payment",
  "request.duration_ms": 145.3
}

The processor creates a histogram data point log.request.duration{service.name="payment"} 145.3.

Value metric

The Value metric processor reads a numeric value from a record field and records it as a gauge data point. Use it to track current states or absolute measurements such as memory usage percentages, temperature readings, or connection counts.

Parameters

Example: Extract memory usage from log records

The following example extracts memory usage percentages from log records and tracks them as a metric.

The processor applies to all records that match the following condition:

isNotNull(memory.usage_percent) AND matchesValue(service.name, "web-*")

The processor is configured as follows:

Unprocessed

{
  "timestamp": "2026-06-01T10:00:00Z",
  "service.name": "web-frontend",
  "memory.usage_percent": 78.4,
  "content": "Memory usage alert"
}

The processor creates a gauge data point log.memory.usage{service.name="web-frontend"} 78.4.

Sampling aware settings Spans

The Sampling Aware Counter Metric, Sampling Aware Histogram Metric, and Sampling Aware Value Metric processors are span-specific variants of their counterparts. They support the same parameters as their counterparts, with the following additions.

Span aggregation and sampling awareness apply to all fields available in field extraction, except duration; duration aggregation is automatically detected and handled.

Configuration best practices

For most production use cases, enable both options to ensure extracted metrics accurately represent total system behavior despite aggregation and sampling optimizations.

Enable Account for aggregated spans when:

Enable Extrapolate metrics to account for trace sampling when:

Disable extrapolation only if you explicitly want metrics that reflect sampled trace data rather than estimated total traffic.

Use cases