feat(health): add primary system UUID and static labels to telemetry

crates/health/Cargo.toml

@@ -89,6 +89,7 @@ nv-redfish = { workspace = true, features = [
 ] }
 rand = { workspace = true }
 url = { workspace = true, features = ["serde"] }
+uuid = { workspace = true }
 x509-parser = { workspace = true }
 
 # [local-dependencies]

crates/health/benches/collector_pipeline.rs

@@ -21,7 +21,7 @@ use std::net::{IpAddr, Ipv4Addr};
 use std::str::FromStr;
 use std::sync::Arc;
 
-use carbide_health::endpoint::{BmcAddr, EndpointMetadata, MachineData};
+use carbide_health::endpoint::{BmcAddr, EndpointMetadata, MachineData, SharedSystemUuid};
 use carbide_health::metrics::MetricsManager;
 use carbide_health::sink::{
     CollectorEvent, CompositeDataSink, DataSink, EventContext, FirmwareInfo, LogRecord,
@@ -59,9 +59,11 @@ fn event_context() -> EventContext {
             mac: MacAddress::from_str("42:9e:b1:bd:9d:dd").unwrap(),
         },
         collector_type: "sensor_collector",
+        labels: Default::default(),
         metadata: Some(EndpointMetadata::Machine(MachineData {
             machine_id: Some(MACHINE_ID.parse().expect("valid machine id")),
             machine_serial: None,
+            system_uuid: SharedSystemUuid::default(),
             slot_number: None,
             tray_index: None,
             nvlink_domain_uuid: None,

crates/health/benches/processor_pipeline.rs

@@ -20,7 +20,7 @@ use std::net::{IpAddr, Ipv4Addr};
 use std::str::FromStr;
 use std::sync::Arc;
 
-use carbide_health::endpoint::{BmcAddr, EndpointMetadata, MachineData};
+use carbide_health::endpoint::{BmcAddr, EndpointMetadata, MachineData, SharedSystemUuid};
 use carbide_health::metrics::MetricsManager;
 use carbide_health::processor::{
     EventProcessingPipeline, EventProcessor, HealthReportProcessor, LeakEventProcessor,
@@ -95,9 +95,11 @@ fn event_context() -> EventContext {
             mac: MacAddress::from_str("42:9e:b1:bd:9d:dd").unwrap(),
         },
         collector_type: "sensor_collector",
+        labels: Default::default(),
         metadata: Some(EndpointMetadata::Machine(MachineData {
             machine_id: Some(MACHINE_ID.parse().expect("valid machine id")),
             machine_serial: None,
+            system_uuid: SharedSystemUuid::default(),
             slot_number: None,
             tray_index: None,
             nvlink_domain_uuid: None,
@@ -273,9 +275,11 @@ fn rack_event_contexts(rack_id: &str, tray_count: usize) -> Vec<EventContext> {
                     mac: MacAddress::from_str(&mac).unwrap(),
                 },
                 collector_type: "sensor_collector",
+                labels: Default::default(),
                 metadata: Some(EndpointMetadata::Machine(MachineData {
                     machine_id: Some(MACHINE_ID.parse().expect("valid machine id")),
                     machine_serial: None,
+                    system_uuid: SharedSystemUuid::default(),
                     slot_number: None,
                     tray_index: None,
                     nvlink_domain_uuid: None,

crates/health/benches/sink_pipeline.rs

@@ -21,7 +21,7 @@ use std::net::{IpAddr, Ipv4Addr};
 use std::str::FromStr;
 use std::sync::Arc;
 
-use carbide_health::endpoint::{BmcAddr, EndpointMetadata, MachineData};
+use carbide_health::endpoint::{BmcAddr, EndpointMetadata, MachineData, SharedSystemUuid};
 use carbide_health::metrics::MetricsManager;
 use carbide_health::sink::{
     Classification, CollectorEvent, CompositeDataSink, DataSink, EventContext, HealthReport,
@@ -69,9 +69,11 @@ fn event_context_for_machine(machine_id: &str) -> EventContext {
             mac: MacAddress::from_str("42:9e:b1:bd:9d:dd").unwrap(),
         },
         collector_type: "sensor_collector",
+        labels: Default::default(),
         metadata: Some(EndpointMetadata::Machine(MachineData {
             machine_id: Some(machine_id.parse().expect("valid machine id")),
             machine_serial: None,
+            system_uuid: SharedSystemUuid::default(),
             slot_number: None,
             tray_index: None,
             nvlink_domain_uuid: None,

crates/health/example/config.example.toml

@@ -40,6 +40,7 @@ port = 443
 mac = "aa:bb:cc:dd:ee:ff"
 username = "admin"
 password = "secret"
+labels = { site = "dev", cluster = "cluster-01", environment = "development" }
 machine = { id = "fm100htjtiaehv1n5vh67tbmqq4eabcjdng40f7jupsadbedhruh6rag1l0", serial = "MN-001", slot_number = 15, tray_index = 5, nvlink_domain_uuid = "00000000-0000-0000-0000-000000000000" }
 
 [[endpoint_sources.static_bmc_endpoints]]

crates/health/src/api_client.rs

@@ -39,7 +39,7 @@ use crate::HealthError;
 use crate::bmc::{BmcClient, BoxFuture, CredentialProvider};
 use crate::endpoint::{
     BmcAddr, BmcCredentials, BmcEndpoint, EndpointMetadata, EndpointSource, MachineData,
-    PowerShelfData, SwitchData, SwitchEndpointRole,
+    PowerShelfData, SharedSystemUuid, SwitchData, SwitchEndpointRole,
 };
 
 /// [`ApiEndpointSource`].
@@ -290,9 +290,11 @@ pub struct ApiEndpointSource {
     bmc_client_cache: Mutex<HashMap<MacAddress, CachedBmcClient>>,
 }
 
+#[derive(Clone)]
 struct CachedBmcClient {
     client: Arc<BmcClient>,
     kind: ApiCredentialKind,
+    system_uuid: SharedSystemUuid,
 }
 
 impl ApiEndpointSource {
@@ -484,6 +486,7 @@ impl ApiEndpointSource {
                     .as_ref()
                     .and_then(|info| info.dmi_data.as_ref())
                     .map(|dmi| dmi.chassis_serial.clone()),
+                system_uuid: SharedSystemUuid::default(),
                 slot_number: machine
                     .placement_in_rack
                     .as_ref()
@@ -588,7 +591,7 @@ impl ApiEndpointSource {
         rack_id: Option<RackId>,
         credential_kind: ApiCredentialKind,
     ) -> Result<Arc<BmcEndpoint>, HealthError> {
-        let bmc = {
+        let cached = {
             let mut cache = self.bmc_client_cache.lock().expect("cache mutex poisoned");
             cache_or_create_bmc_client(&mut cache, addr.mac, credential_kind, |kind| {
                 let provider: Arc<dyn CredentialProvider> = Arc::new(ApiCredentialProvider {
@@ -604,11 +607,16 @@ impl ApiEndpointSource {
                 )?))
             })?
         };
+        let mut metadata = metadata;
+        if let Some(EndpointMetadata::Machine(machine)) = metadata.as_mut() {
+            machine.system_uuid = cached.system_uuid;
+        }
         Ok(Arc::new(BmcEndpoint {
             addr,
             metadata,
             rack_id,
-            bmc,
+            labels: Default::default(),
+            bmc: cached.client,
         }))
     }
 }
@@ -618,7 +626,7 @@ fn cache_or_create_bmc_client(
     mac: MacAddress,
     credential_kind: ApiCredentialKind,
     make_client: impl FnOnce(ApiCredentialKind) -> Result<Arc<BmcClient>, HealthError>,
-) -> Result<Arc<BmcClient>, HealthError> {
+) -> Result<CachedBmcClient, HealthError> {
     if let Some(existing) = cache.get(&mac) {
         if existing.kind != credential_kind {
             return Err(HealthError::GenericError(format!(
@@ -629,18 +637,17 @@ fn cache_or_create_bmc_client(
                 credential_kind.tag(),
             )));
         }
-        return Ok(existing.client.clone());
+        return Ok(existing.clone());
     }
 
     let client = make_client(credential_kind.clone())?;
-    cache.insert(
-        mac,
-        CachedBmcClient {
-            client: client.clone(),
-            kind: credential_kind,
-        },
-    );
-    Ok(client)
+    let cached = CachedBmcClient {
+        client,
+        kind: credential_kind,
+        system_uuid: SharedSystemUuid::default(),
+    };
+    cache.insert(mac, cached.clone());
+    Ok(cached)
 }
 
 /// Returns the machine-level GPU driver version derived from discovery data.
@@ -840,8 +847,8 @@ mod tests {
         );
     }
 
-    #[test]
-    fn cache_returns_existing_client_on_matching_kind() {
+    #[tokio::test]
+    async fn cache_returns_existing_client_on_matching_kind() {
         let mut cache: HashMap<MacAddress, CachedBmcClient> = HashMap::new();
         let factory_calls = AtomicUsize::new(0);
 
@@ -860,10 +867,21 @@ mod tests {
             .expect("cache hit");
 
         assert!(
-            Arc::ptr_eq(&first, &second),
+            Arc::ptr_eq(&first.client, &second.client),
             "cache hit must reuse the same BmcClient Arc — otherwise every \
              iteration of discovery rebuilds the session and re-fetches creds"
         );
+        let system_uuid = uuid::uuid!("4c4c4544-0044-4710-8052-cac04f4b4632");
+        first
+            .system_uuid
+            .get_or_try_init(|| async { Ok::<_, std::convert::Infallible>(Some(system_uuid)) })
+            .await
+            .expect("infallible UUID initialization");
+        assert_eq!(
+            second.system_uuid.get(),
+            Some(system_uuid),
+            "cache hit must reuse machine UUID state across discovery iterations"
+        );
         assert_eq!(
             factory_calls.load(Ordering::SeqCst),
             1,

crates/health/src/collectors/leak_detector.rs

@@ -271,7 +271,7 @@ mod tests {
     use mac_address::MacAddress;
 
     use super::*;
-    use crate::endpoint::{BmcAddr, EndpointMetadata, MachineData};
+    use crate::endpoint::{BmcAddr, EndpointMetadata, MachineData, SharedSystemUuid};
     use crate::sink::HealthReportTarget;
 
     #[derive(Debug, Eq, PartialEq)]
@@ -337,12 +337,14 @@ mod tests {
             metadata: Some(EndpointMetadata::Machine(MachineData {
                 machine_id: None,
                 machine_serial: None,
+                system_uuid: SharedSystemUuid::default(),
                 slot_number: None,
                 tray_index: None,
                 nvlink_domain_uuid: None,
                 driver_version: None,
             })),
             rack_id: None,
+            labels: Default::default(),
         }
     }
 

crates/health/src/collectors/nvue/gnmi/on_change_processor.rs

@@ -383,6 +383,7 @@ mod tests {
             collector_type,
             metadata: None,
             rack_id: None,
+            labels: Default::default(),
         }
     }
 
@@ -779,6 +780,7 @@ mod tests {
                     mac: MacAddress::from_str("AA:BB:CC:DD:EE:FF").unwrap(),
                 },
                 collector_type: ON_CHANGE_STREAM_ID_SYSTEM_EVENTS,
+                labels: Default::default(),
                 metadata: Some(EndpointMetadata::Switch(SwitchData {
                     id: Some(switch_id),
                     serial: "SN-SWITCH-001".to_string(),

crates/health/src/collectors/nvue/gnmi/sample_processor.rs

@@ -1102,6 +1102,7 @@ mod tests {
             collector_type: NVUE_GNMI_SAMPLE_STREAM_ID,
             metadata: None,
             rack_id: None,
+            labels: Default::default(),
         };
         GnmiSampleProcessor {
             data_sink: None,
@@ -1167,6 +1168,7 @@ mod tests {
                     mac: MacAddress::from_str("AA:BB:CC:DD:EE:FF").unwrap(),
                 },
                 collector_type: NVUE_GNMI_SAMPLE_STREAM_ID,
+                labels: Default::default(),
                 metadata: Some(EndpointMetadata::Switch(SwitchData {
                     id: Some(switch_id),
                     serial: "SN-SWITCH-001".to_string(),

crates/health/src/collectors/nvue/rest/collector.rs

@@ -1260,6 +1260,7 @@ mod tests {
             collector_type: COLLECTOR_NAME,
             metadata: None,
             rack_id: None,
+            labels: Default::default(),
         };
 
         NvueRestCollector {