feat(spans): extract cache size and cache hit in indexed spans

CHANGELOG.md

@@ -1,6 +1,6 @@
 # Changelog
 
 ## Unreleased
 
 **Features**:
 
@@ -13,6 +13,7 @@
 
 **Internal**:
 
+- Extract `cache.item_size` and `cache.hit` data into span indexed ([#3367]https://github.com/getsentry/relay/pull/3367)
 - Enable `db.redis` span metrics extraction. ([#3283](https://github.com/getsentry/relay/pull/3283))
 - Add data categories for continuous profiling. ([#3284](https://github.com/getsentry/relay/pull/3284), [#3303](https://github.com/getsentry/relay/pull/3303))
 - Apply rate limits to span metrics. ([#3255](https://github.com/getsentry/relay/pull/3255))

relay-event-normalization/src/normalize/span/tag_extraction.rs

@@ -66,6 +66,8 @@ pub enum SpanTagKey {
     /// The start type of the application when the span occurred.
     AppStartType,
     ReplayId,
+    CacheHit,
+    CacheItemSize,
 }
 
 impl SpanTagKey {
@@ -104,6 +106,7 @@ impl SpanTagKey {
             SpanTagKey::TimeToInitialDisplay => "ttid",
             SpanTagKey::FileExtension => "file_extension",
             SpanTagKey::MainThread => "main_thread",
+            SpanTagKey::CacheHit => "cache.hit",
             SpanTagKey::OsName => "os.name",
             SpanTagKey::AppStartType => "app_start_type",
             SpanTagKey::ReplayId => "replay_id",
@@ -406,6 +409,13 @@ pub fn extract_tags(
             }
         }
 
+        if span_op.starts_with("cache.") {
+            if let Some(Value::Bool(cache_hit)) = span.data.value().and_then(|data| data.cache_hit.value()) {
+                let tag_value = if cache_hit { "true" } else { "false" };
+                span_tags.insert(SpanTagKey::CacheHit, tag_value.to_owned());
+            }
+        }
+
         if let Some(scrubbed_desc) = scrubbed_description {
             // Truncating the span description's tag value is, for now,
             // a temporary solution to not get large descriptions dropped. The
@@ -559,6 +569,27 @@ pub fn extract_measurements(span: &mut Span) {
         return;
     };
 
+    if span_op.starts_with("cache.") {
+        if let Some(data) = span.data.value() {
+            if let Some(value) = match &data.cache_item_size.value() {
+                Some(Value::F64(f)) => Some(*f),
+                Some(Value::I64(i)) => Some(*i as f64),
+                Some(Value::U64(u)) => Some(*u as f64),
+                _ => None,
+            } {
+                let measurements = span.measurements.get_or_insert_with(Default::default);
+                measurements.insert(
+                    "cache.item_size".to_owned(),
+                    Measurement {
+                        value: value.into(),
+                        unit: MetricUnit::Information(InformationUnit::Byte).into(),
+                    }
+                    .into(),
+                );
+            }
+        }
+    }
+
     if span_op.starts_with("resource.") {
         if let Some(data) = span.data.value() {
             for (field, key) in [
@@ -1283,6 +1314,81 @@ LIMIT 1
         assert!(!tags_3.contains_key("raw_domain"));
     }
 
+    #[test]
+    fn test_cache_extraction() {
+        let json = r#"
+            {
+                "spans": [
+                    {
+                        "timestamp": 1694732408.3145,
+                        "start_timestamp": 1694732407.8367,
+                        "exclusive_time": 477.800131,
+                        "description": "get my_key",
+                        "op": "cache.get_item",
+                        "span_id": "97c0ef9770a02f9d",
+                        "parent_span_id": "9756d8d7b2b364ff",
+                        "trace_id": "77aeb1c16bb544a4a39b8d42944947a3",
+                        "data": {
+                            "cache.hit": true,
+                            "cache.item_size": 8,
+                            "thread.id": "6286962688",
+                            "thread.name": "Thread-4 (process_request_thread)"
+
+                        },
+                        "hash": "e2fae740cccd3781"
+                    },
+                    {
+                        "timestamp": 1694732409.3145,
+                        "start_timestamp": 1694732408.8367,
+                        "exclusive_time": 477.800131,
+                        "description": "get my_key",
+                        "op": "cache.get_item",
+                        "span_id": "97c0ef9770a02f9d",
+                        "parent_span_id": "9756d8d7b2b364ff",
+                        "trace_id": "77aeb1c16bb544a4a39b8d42944947a3",
+                        "data": {
+                            "cache.hit": false,
+                            "cache.item_size": 8,
+                            "thread.id": "6286962688",
+                            "thread.name": "Thread-4 (process_request_thread)"
+
+                        },
+                        "hash": "e2fae740cccd3781"
+                    }
+                ]
+            }
+        "#;
+
+        let mut event = Annotated::<Event>::from_json(json)
+            .unwrap()
+            .into_value()
+            .unwrap();
+
+        extract_span_tags_from_event(&mut event, 200);
+
+        let span_1 = &event.spans.value().unwrap()[0];
+        let span_2 = &event.spans.value().unwrap()[1];
+
+        let tags_1 = get_value!(span_1.sentry_tags).unwrap();
+        let tags_2 = get_value!(span_2.sentry_tags).unwrap();
+        let measurements_1 = span_1.value().unwrap().measurements.value().unwrap();
+
+        assert_eq!(tags_1.get("cache.hit").unwrap().as_str(), Some("true"));
+        assert_eq!(tags_2.get("cache.hit").unwrap().as_str(), Some("false"));
+        assert_debug_snapshot!(measurements_1, @r###"
+        Measurements(
+            {
+                "cache.item_size": Measurement {
+                    value: 8.0,
+                    unit: Information(
+                        Byte,
+                    ),
+                },
+            },
+        )
+        "###);
+    }
+
     #[test]
     fn test_mobile_specific_tags() {
         let json = r#"

relay-event-schema/src/protocol/span.rs

@@ -231,10 +231,18 @@ pub struct SpanData {
     #[metastructure(field = "resource.render_blocking_status")]
     pub resource_render_blocking_status: Annotated<Value>,
 
-    /// Name of the database host.
+    /// Name of the web server host.
     #[metastructure(field = "server.address")]
     pub server_address: Annotated<Value>,
 
+    /// Whether cache was hit or miss on a read operation.
+    #[metastructure(field = "cache.hit")]
+    pub cache_hit: Annotated<Value>,
+
+    /// The size of the cache item.
+    #[metastructure(field = "cache.item_size")]
+    pub cache_item_size: Annotated<Value>,
+
     /// The status HTTP response.
     #[metastructure(field = "http.response.status_code", legacy_alias = "status_code")]
     pub http_response_status_code: Annotated<Value>,
@@ -512,6 +520,8 @@ mod tests {
             http_response_transfer_size: ~,
             resource_render_blocking_status: ~,
             server_address: ~,
+            cache_hit: ~,
+            cache_item_size: ~,
             http_response_status_code: ~,
             ai_input_messages: ~,
             ai_completion_tokens_used: ~,

relay-server/src/metrics_extraction/snapshots/relay_server__metrics_extraction__event__tests__extract_span_metrics_mobile.snap

@@ -324,6 +324,8 @@ expression: "(&event.value().unwrap().spans, metrics)"
                 http_response_transfer_size: ~,
                 resource_render_blocking_status: ~,
                 server_address: ~,
+                cache_hit: ~,
+                cache_item_size: ~,
                 http_response_status_code: ~,
                 ai_input_messages: ~,
                 ai_completion_tokens_used: ~,
@@ -403,6 +405,8 @@ expression: "(&event.value().unwrap().spans, metrics)"
                 http_response_transfer_size: ~,
                 resource_render_blocking_status: ~,
                 server_address: ~,
+                cache_hit: ~,
+                cache_item_size: ~,
                 http_response_status_code: ~,
                 ai_input_messages: ~,
                 ai_completion_tokens_used: ~,