The aggregation params of metrics can be defined #844

CHANGELOG.md

@@ -2,6 +2,11 @@
 
 All notable changes to this project will be documented in this file.
 
+# [Next] - Next
+
+- Support for describing aggregation being done on metrics
+  ([#845](https://github.com/open-telemetry/weaver/pull/822) by @thompson-tomo)
+
 # [0.17.1] - 2025-08-15
 
 - Fix error messages to ignore new version variants ([#880](https://github.com/open-telemetry/weaver/pull/880) by @jsuereth)

crates/weaver_forge/src/registry.rs

@@ -12,6 +12,7 @@ use weaver_resolved_schema::attribute::Attribute;
 use weaver_resolved_schema::catalog::Catalog;
 use weaver_resolved_schema::lineage::GroupLineage;
 use weaver_resolved_schema::registry::{Group, Registry};
+use weaver_semconv::aggregation::AggregationSpec;
 use weaver_semconv::any_value::AnyValueSpec;
 use weaver_semconv::deprecated::Deprecated;
 use weaver_semconv::group::{GroupType, InstrumentSpec, SpanKindSpec};
@@ -99,6 +100,12 @@ pub struct ResolvedGroup {
     /// Note: This field is required if type is metric.
     #[serde(skip_serializing_if = "Option::is_none")]
     pub unit: Option<String>,
+    /// The aggregation which should occur on the data points being capture by a meter.
+    /// Semconv metrics all use the default aggregation type, hence this option is for 
+    /// providing the parameters of the aggregation.
+    /// For more details: [Metrics SDK - Aggregation](https://opentelemetry.io/docs/specs/otel/metrics/sdk/#aggregation).
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub aggregation: Option<AggregationSpec>,
     /// The name of the event. If not specified, the prefix is used.
     /// If prefix is empty (or unspecified), name is required.
     #[serde(skip_serializing_if = "Option::is_none")]
@@ -168,6 +175,7 @@ impl ResolvedGroup {
             metric_name: group.metric_name.clone(),
             instrument: group.instrument.clone(),
             unit: group.unit.clone(),
+            aggregation: group.aggregation.clone(),
             name: group.name.clone(),
             lineage,
             display_name: group.display_name.clone(),
@@ -229,6 +237,7 @@ impl ResolvedRegistry {
                     metric_name: group.metric_name.clone(),
                     instrument: group.instrument.clone(),
                     unit: group.unit.clone(),
+                    aggregation: group.aggregation.clone(),
                     name: group.name.clone(),
                     lineage,
                     display_name: group.display_name.clone(),

crates/weaver_resolved_schema/src/registry.rs

@@ -16,6 +16,7 @@ use crate::registry::GroupStats::{
     AttributeGroup, Entity, Event, Metric, MetricGroup, Scope, Span, Undefined,
 };
 use serde::{Deserialize, Serialize};
+use weaver_semconv::aggregation::AggregationSpec;
 use weaver_semconv::deprecated::Deprecated;
 use weaver_semconv::group::{GroupType, InstrumentSpec, SpanKindSpec};
 use weaver_semconv::provenance::Provenance;
@@ -113,6 +114,12 @@ pub struct Group {
     /// Note: This field is required if type is metric.
     #[serde(skip_serializing_if = "Option::is_none")]
     pub unit: Option<String>,
+    /// The aggregation which should occur on the data points being capture by a meter.
+    /// Semconv metrics all use the default aggregation type, hence this option is for 
+    /// providing the parameters of the aggregation.
+    /// For more details: [Metrics SDK - Aggregation](https://opentelemetry.io/docs/specs/otel/metrics/sdk/#aggregation).
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub aggregation: Option<AggregationSpec>,
     /// The name of the event. If not specified, the prefix is used.
     /// If prefix is empty (or unspecified), name is required.
     #[serde(skip_serializing_if = "Option::is_none")]

crates/weaver_resolver/src/registry.rs

@@ -413,6 +413,7 @@ fn group_from_spec(group: GroupSpecWithProvenance) -> UnresolvedGroup {
             metric_name: group.spec.metric_name,
             instrument: group.spec.instrument,
             unit: group.spec.unit,
+            aggregation: group.spec.aggregation,
             name: group.spec.name,
             lineage: Some(GroupLineage::new(group.provenance.clone())),
             display_name: group.spec.display_name,

crates/weaver_semconv/data/jvm-metrics.yaml

@@ -71,6 +71,13 @@ groups:
     brief: "Duration of JVM garbage collection actions."
     instrument: histogram
     unit: "s"
+    aggregation:
+      parameters:
+        boundaries:
+          - 0
+          - 5
+          - 10
+        recordMinMax: true
     attributes:
       - id: jvm.gc.name
         stability: stable

crates/weaver_semconv/src/aggregation.rs

@@ -0,0 +1,46 @@
+// SPDX-License-Identifier: Apache-2.0
+
+//! Metric specification.
+//! 
+use schemars::JsonSchema;
+use serde::{Deserialize, Serialize};
+use std::collections::HashMap;
+use crate::YamlValue;
+
+/// An aggregation specification.
+#[derive(Serialize, Deserialize, Clone, Debug, Eq, PartialEq, JsonSchema)]
+#[serde(deny_unknown_fields)]
+pub struct AggregationSpec {
+    /// The method of aggregation to be used
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub method: Option<AggregationMethodSpec>,
+    /// The parameters used in the aggregation
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub parameters: Option<HashMap<String, YamlValue>>,
+}
+
+impl AggregationSpec {
+    /// Returns the parameters of the aggregation.
+    #[must_use]
+    pub fn parameters(&self) -> &Option<HashMap<String, YamlValue>> {
+        &self.parameters
+    }
+}
+
+/// The Aggregation Method
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, Eq, Hash, JsonSchema)]
+#[serde(rename_all = "snake_case")]
+pub enum AggregationMethodSpec {
+    /// Use the Default Aggregation Method.
+    Default,
+    /// Use the Sum Aggregation Method.
+    Sum,
+    /// Use the Last Value Aggregation Method.
+    LastValue,
+    /// Use the Histogram Aggregation Method.
+    Histogram,
+    /// Use the Explicit Histogram Aggregation Method.
+    ExplicitHistogram,
+    /// Use the Exponential Histogram Aggregation Method.
+    ExponentialHistogram,
+}

crates/weaver_semconv/src/group.rs

@@ -10,6 +10,7 @@ use serde::{Deserialize, Serialize};
 use std::collections::{BTreeMap, HashSet};
 use std::fmt::{Display, Formatter};
 
+use crate::aggregation::AggregationSpec;
 use crate::any_value::AnyValueSpec;
 use crate::attribute::{AttributeSpec, AttributeType, PrimitiveOrArrayTypeSpec};
 use crate::deprecated::Deprecated;
@@ -97,6 +98,12 @@ pub struct GroupSpec {
     /// Note: This field is required if type is metric.
     #[serde(skip_serializing_if = "Option::is_none")]
     pub unit: Option<String>,
+    /// The aggregation which should occur on the data points being capture by a meter.
+    /// Semconv metrics all use the default aggregation type, hence this option is for 
+    /// providing the parameters of the aggregation.
+    /// For more details: [Metrics SDK - Aggregation](https://opentelemetry.io/docs/specs/otel/metrics/sdk/#aggregation).
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub aggregation: Option<AggregationSpec>,
     /// The name of the event (valid only when the group `type` is `event`).
     ///
     /// Note: If not specified, the prefix is used. If the prefix is empty (or unspecified), the name is required.
@@ -694,6 +701,7 @@ mod tests {
             metric_name: None,
             instrument: None,
             unit: None,
+            aggregation: None,
             name: None,
             display_name: None,
             body: None,
@@ -859,6 +867,7 @@ mod tests {
             metric_name: None,
             instrument: None,
             unit: None,
+            aggregation: None,
             name: None,
             display_name: None,
             body: None,
@@ -1145,6 +1154,7 @@ mod tests {
             metric_name: None,
             instrument: None,
             unit: None,
+            aggregation: None,
             display_name: None,
             attributes: vec![],
             body: Some(AnyValueSpec::String {
@@ -1360,6 +1370,7 @@ mod tests {
             metric_name: None,
             instrument: None,
             unit: None,
+            aggregation: None,
             display_name: None,
             attributes: vec![],
             body: Some(AnyValueSpec::String {
@@ -1515,6 +1526,7 @@ mod tests {
             metric_name: None,
             instrument: None,
             unit: None,
+            aggregation: None,
             name: None,
             display_name: None,
             body: None,
@@ -1586,6 +1598,7 @@ mod tests {
         group.metric_name = Some("test".to_owned());
         group.instrument = Some(Counter);
         group.unit = Some("test".to_owned());
+        group.aggregation = None;
         let result = group.validate("<test>").into_result_failing_non_fatal();
         assert_eq!(
             Err(InvalidGroupStability {
@@ -1685,6 +1698,7 @@ mod tests {
             metric_name: None,
             instrument: None,
             unit: None,
+            aggregation: None,
             name: None,
             display_name: None,
             body: None,
@@ -1776,6 +1790,7 @@ mod tests {
         group.metric_name = Some("test".to_owned());
         group.instrument = Some(Counter);
         group.unit = Some("test".to_owned());
+        group.aggregation = None;
         assert!(group
             .validate("<test>")
             .into_result_failing_non_fatal()
@@ -1837,6 +1852,7 @@ mod tests {
             metric_name: None,
             instrument: None,
             unit: None,
+            aggregation: None,
             name: None,
             display_name: None,
             body: None,
@@ -1897,6 +1913,7 @@ mod tests {
             metric_name: Some("metric".to_owned()),
             instrument: Some(Gauge),
             unit: Some("{thing}".to_owned()),
+            aggregation: None,
             name: None,
             display_name: None,
             body: None,
@@ -1913,6 +1930,7 @@ mod tests {
         group.metric_name = None;
         group.instrument = None;
         group.unit = None;
+        group.aggregation = None;
         group.span_kind = Some(SpanKindSpec::Client);
         assert!(group
             .validate("<test>")

crates/weaver_semconv/src/lib.rs

@@ -13,6 +13,7 @@ use std::path::PathBuf;
 use weaver_common::diagnostic::{DiagnosticMessage, DiagnosticMessages};
 use weaver_common::error::{format_errors, WeaverError};
 
+pub mod aggregation;
 pub mod any_value;
 pub mod attribute;
 pub mod deprecated;

crates/weaver_semconv/src/registry.rs

@@ -339,6 +339,7 @@ mod tests {
                         metric_name: None,
                         instrument: None,
                         unit: None,
+                        aggregation: None,
                         brief: "brief".to_owned(),
                         note: "note".to_owned(),
                         extends: None,
@@ -366,6 +367,7 @@ mod tests {
                         metric_name: None,
                         instrument: None,
                         unit: None,
+                        aggregation: None,
                         brief: "brief".to_owned(),
                         note: "note".to_owned(),
                         extends: None,