fix(metrics): align Prometheus metric names and units with conventions

- `nodedb_bridge_utilization` (integer percent) becomes
  `nodedb_bridge_utilization_ratio` (float 0.0–1.0).
- `nodedb_compaction_throughput_bytes_sec` becomes
  `nodedb_compaction_throughput_bytes_per_second`.
- `nodedb_mmap_major_faults` becomes `nodedb_mmap_major_faults_total`
  (counter suffix required by Prometheus naming conventions).
- `shutdown_last_duration_ms{phase}` becomes
  `nodedb_shutdown_phase_duration_seconds{phase}` with millisecond values
  converted to fractional seconds.
- `catalog_sanity_check_total` becomes `nodedb_catalog_sanity_check_total`.

All references in `shutdown/bus.rs` and `main.rs` are updated to the new
metric names.

Author: farhan-syah

nodedb/src/control/metrics/prometheus/core.rs

@@ -75,11 +75,11 @@ impl SystemMetrics {
         );
 
         // ── Bridge ──
-        gauge(
+        gauge_f64(
             out,
-            "nodedb_bridge_utilization",
-            "SPSC bridge utilization percent",
-            self.bridge_utilization.load(Ordering::Relaxed),
+            "nodedb_bridge_utilization_ratio",
+            "SPSC bridge utilization as a ratio (0.0–1.0)",
+            self.bridge_utilization.load(Ordering::Relaxed) as f64 / 100.0,
         );
 
         // ── Compaction ──
@@ -97,8 +97,8 @@ impl SystemMetrics {
         );
         gauge(
             out,
-            "nodedb_compaction_throughput_bytes_sec",
-            "Compaction throughput bytes/sec",
+            "nodedb_compaction_throughput_bytes_per_second",
+            "Compaction throughput bytes per second",
             self.compaction_throughput_bytes_sec.load(Ordering::Relaxed),
         );
 

nodedb/src/control/metrics/prometheus/engines.rs

@@ -189,7 +189,7 @@ impl SystemMetrics {
         // ── Contention ──
         counter(
             out,
-            "nodedb_mmap_major_faults",
+            "nodedb_mmap_major_faults_total",
             "mmap major page faults",
             self.mmap_major_faults.load(Ordering::Relaxed),
         );

nodedb/src/control/metrics/system.rs

@@ -489,7 +489,7 @@ impl SystemMetrics {
         out
     }
 
-    /// Emit `shutdown_last_duration_ms{phase}` gauges.
+    /// Emit `nodedb_shutdown_phase_duration_seconds{phase}` gauges.
     fn prometheus_shutdown_phases(&self, out: &mut String) {
         use std::fmt::Write as _;
         let m = self
@@ -500,17 +500,21 @@ impl SystemMetrics {
             return;
         }
         let _ = out.write_str(
-            "# HELP shutdown_last_duration_ms Duration of each shutdown phase in the last graceful shutdown\n\
-             # TYPE shutdown_last_duration_ms gauge\n",
+            "# HELP nodedb_shutdown_phase_duration_seconds Duration of each shutdown phase in the last graceful shutdown\n\
+             # TYPE nodedb_shutdown_phase_duration_seconds gauge\n",
         );
         let mut pairs: Vec<_> = m.iter().collect();
         pairs.sort_by(|a, b| a.0.cmp(b.0));
         for (phase, ms) in pairs {
-            let _ = writeln!(out, r#"shutdown_last_duration_ms{{phase="{phase}"}} {ms}"#);
+            let secs = *ms as f64 / 1_000.0;
+            let _ = writeln!(
+                out,
+                r#"nodedb_shutdown_phase_duration_seconds{{phase="{phase}"}} {secs}"#
+            );
         }
     }
 
-    /// Emit `catalog_sanity_check_total{registry,outcome}` labeled counters.
+    /// Emit `nodedb_catalog_sanity_check_total{registry,outcome}` labeled counters.
     fn prometheus_catalog_sanity(&self, out: &mut String) {
         use std::fmt::Write as _;
         let m = self
@@ -521,15 +525,15 @@ impl SystemMetrics {
             return;
         }
         let _ = out.write_str(
-            "# HELP catalog_sanity_check_total Catalog sanity check outcomes per registry\n\
-             # TYPE catalog_sanity_check_total counter\n",
+            "# HELP nodedb_catalog_sanity_check_total Catalog sanity check outcomes per registry\n\
+             # TYPE nodedb_catalog_sanity_check_total counter\n",
         );
         let mut pairs: Vec<_> = m.iter().collect();
         pairs.sort_by(|a, b| a.0.cmp(b.0));
         for ((registry, outcome), count) in pairs {
             let _ = writeln!(
                 out,
-                r#"catalog_sanity_check_total{{registry="{registry}",outcome="{outcome}"}} {count}"#
+                r#"nodedb_catalog_sanity_check_total{{registry="{registry}",outcome="{outcome}"}} {count}"#
             );
         }
     }

nodedb/src/control/otel/exporter.rs

@@ -129,13 +129,13 @@ fn build_metrics_request(metrics: &SystemMetrics, node_id: u64) -> ExportMetrics
             now_ns,
         ),
         gauge_metric(
-            "nodedb_wal_fsync_latency_us",
-            metrics.wal_fsync_latency_micros.load(Ordering::Relaxed) as f64,
+            "nodedb_wal_fsync_latency_seconds",
+            metrics.wal_fsync_latency_micros.load(Ordering::Relaxed) as f64 / 1_000_000.0,
             now_ns,
         ),
         gauge_metric(
-            "nodedb_bridge_utilization",
-            metrics.bridge_utilization.load(Ordering::Relaxed) as f64,
+            "nodedb_bridge_utilization_ratio",
+            metrics.bridge_utilization.load(Ordering::Relaxed) as f64 / 100.0,
             now_ns,
         ),
         gauge_metric(

nodedb/src/control/shutdown/bus.rs

@@ -322,7 +322,7 @@ impl ShutdownBus {
         *self.phase_tx.borrow()
     }
 
-    /// Wire a metrics sink so the bus records `shutdown_last_duration_ms{phase}`
+    /// Wire a metrics sink so the bus records `nodedb_shutdown_phase_duration_seconds{phase}`
     /// for each phase transition during shutdown.
     ///
     /// Must be called before `initiate()` to have effect. Idempotent.

nodedb/src/main.rs

@@ -640,7 +640,7 @@ async fn main() -> anyhow::Result<()> {
     // existing `watch::Receiver<bool>` subscribers wake up as well.
     let (shutdown_bus, _shutdown_bus_handle) =
         nodedb::control::shutdown::ShutdownBus::new(Arc::clone(&shared.shutdown));
-    // Wire system metrics so the bus records `shutdown_last_duration_ms{phase}`
+    // Wire system metrics so the bus records `nodedb_shutdown_phase_duration_seconds{phase}`
     // for each phase transition during graceful shutdown.
     shutdown_bus.set_metrics(Arc::clone(&system_metrics));