Signal to Metrics

Description

This connector generates custom metrics from logs, metrics, and traces using OTTL expressions.

Use

Use this connector to:

• Count log events as metrics for monitoring error rates or request frequencies

• Generate latency histograms from trace spans

• Derive new metrics from existing metric datapoints

• Build custom service level indicators from raw telemetry

Stability

• Alpha

Supported Types

Metrics	Logs	Traces
✓	✓	✓

All input types produce metrics as output.

Configuration

Field	Description
Telemetry Types	Select which signal types to process: Logs, Metrics, or Traces. At least one must be selected.
Metric Definitions	One or more metrics to generate. See metric definition fields below.

Metric Definition

Field	Description
Name	The metric name (e.g., http.request.count).
Description	Description of the metric.
Unit	Metric unit (e.g., ms, bytes, logs).
Type	Sum, Gauge, Histogram, or Exponential Histogram.
Value Expression	OTTL expression that returns the metric value.
Condition	OTTL condition to filter which signals generate this metric.
Buckets	Bucket boundaries (Histogram only). Default: [2, 4, 6, 8, 10, 50, 100, 200, 400, 800, 1000, 1400, 2000, 5000, 10000, 15000].
Max Size	Max bucket count (Exponential Histogram only). Default: 160.
Attributes	Attributes to include in the output metric.
Resource Attributes	Resource attributes to include in the output metric.

Metric Types

Type	Description
Sum	Aggregates numeric values. Use for counters and cumulative metrics.
Gauge	Records the last observed value. Use for current state metrics.
Histogram	Distributes values across explicit buckets. Optionally configure Buckets (default: [2, 4, 6, 8, 10, 50, 100, 200, 400, 800, 1000, 1400, 2000, 5000, 10000, 15000]).
Exponential Histogram	Uses dynamic bucket sizing. Optionally configure Max Size (default: 160).

Attributes

Specify which attributes from the source signal to include in the output metric. If no attributes are defined, the output metric will have no attributes.

Field	Description
Key	The attribute key to include.
Behavior	Required (default): incoming signals missing this attribute are skipped entirely. Optional: attribute is included if present on incoming signals, excluded if not. Has Default: use default value if attribute is missing from incoming signals.
Default Value	Value to use when attribute is missing (only with Has Default).

Resource Attributes

Specify which resource attributes to include in the output metric. If no resource attributes are defined, all incoming resource attributes are included.

Field	Description
Key	The resource attribute key to include.
Behavior	Optional (default) or Has Default. Same behavior as attributes, but Required is not available.
Default Value	Value to use when attribute is missing (only with Has Default).

Conditions

Filter which signals generate metrics using OTTL conditions.

Example conditions:

• attributes["http.status_code"] >= 400 — only error responses

• severity_number >= SEVERITY_NUMBER_ERROR — only error logs

• name == "HTTP GET" — specific span names

OTTL Expressions

The connector supports OTTL functions for extracting values.

Common expressions:

• 1 — literal value for counting

• Int(attributes["response_time"]) — extract integer attribute

• Double(attributes["duration_ms"]) — extract float attribute

For spans, AdjustedCount() calculates accurate counts when sampling is enabled:

Int(AdjustedCount())

Example Configuration(s)

Count Log Records by Severity

Count all log records, grouped by severity level.

• Signal Type: logs

• Name: log.record.count

• Type: Sum

• Value Expression: 1

• Attributes: Key severity, Behavior Optional

HTTP Request Duration Histogram

Generate a latency histogram from HTTP server spans.

• Signal Type: spans

• Name: http.server.duration

• Unit: ms

• Type: Histogram

• Value Expression: Double(attributes["http.server.request.duration"])

• Condition: kind == SPAN_KIND_SERVER

• Buckets: [5, 10, 25, 50, 100, 250, 500, 1000, 2500, 5000, 10000]

• Attributes: Key http.route, Behavior Optional; Key http.method, Behavior Required

Error Count by Service

Count error logs grouped by service name.

• Signal Type: logs

• Name: error.count

• Type: Sum

• Value Expression: 1

• Condition: severity_number >= SEVERITY_NUMBER_ERROR

• Resource Attributes: Key service.name, Behavior Optional

Span Count with Sampling Adjustment

Accurately count spans when sampling is enabled.

• Signal Type: spans

• Name: span.count

• Type: Sum

• Value: Int(AdjustedCount())

Technical Notes

The connector automatically adds signaltometrics.service.instance.id to every output metric for single-writer compliance.

The connector does not perform time-based aggregation. Metrics are aggregated per batch only.

Related Resources

• OTTL Syntax Reference

• OTTL Functions

• OTEL Signal to Metrics Documentation