chore(node): restore to check host CPU usage

maidsafe · Jan 14, 2025 · 0d1c4c4 · 0d1c4c4
1 parent dfd692f
commit 0d1c4c4
Show file tree

Hide file tree

Showing 2 changed files with 36 additions and 54 deletions.
diff --git a/ant-logging/src/metrics.rs b/ant-logging/src/metrics.rs
@@ -25,7 +25,7 @@ struct Metrics {
     process: Option<ProcessMetrics>,
 }
 
-#[derive(Clone, Debug, Default, Serialize)]
+#[derive(Debug, Serialize)]
 struct ProcessMetrics {
     // Percentage of CPU used by the process
     cpu_usage_percent: f32,
@@ -73,10 +73,10 @@ pub async fn init_metrics(pid: u32) {
                 }
             };
 
-            // let system_cpu_usage_percent = sys.global_cpu_usage();
+            let system_cpu_usage_percent = sys.global_cpu_usage();
             let metrics = Metrics {
                 physical_cpu_threads: sys.cpus().len(),
-                system_cpu_usage_percent: process.clone().unwrap_or_default().cpu_usage_percent,
+                system_cpu_usage_percent,
                 process,
             };
             match serde_json::to_string(&metrics) {
@@ -91,8 +91,8 @@ pub async fn init_metrics(pid: u32) {
 
 // Refreshes only the metrics that we interested in.
 fn refresh_metrics(sys: &mut System, networks: &mut Networks, pid: Pid) {
-    // sys.refresh_memory();
-    // sys.refresh_cpu_usage();
+    sys.refresh_memory();
+    sys.refresh_cpu_usage();
     networks.refresh(true);
 
     // To refresh only the specific process:

diff --git a/ant-node/src/bin/antnode/main.rs b/ant-node/src/bin/antnode/main.rs
@@ -36,7 +36,7 @@ use std::{
     process::Command,
     time::Duration,
 };
-use sysinfo::{self, Pid, ProcessRefreshKind, ProcessesToUpdate, System};
+use sysinfo::{self, System};
 use tokio::{
     runtime::Runtime,
     sync::{broadcast::error::RecvError, mpsc},
@@ -413,14 +413,12 @@ You can check your reward balance by running:
         }
     });
     let ctrl_tx_clone_cpu = ctrl_tx.clone();
-    let pid = Pid::from_u32(std::process::id());
-    // Monitor process CPU usage
+    // Monitor Host CPU usage
     tokio::spawn(async move {
         use rand::{thread_rng, Rng};
 
         const CPU_CHECK_INTERVAL: Duration = Duration::from_secs(60);
-        // As this is now a per-process CPU usage, it shall never exceeds 10% for too long time.
-        const CPU_USAGE_THRESHOLD: f64 = 10.0;
+        const CPU_USAGE_THRESHOLD: f32 = 50.0;
         const HIGH_CPU_CONSECUTIVE_LIMIT: u8 = 5;
         const NODE_STOP_DELAY: Duration = Duration::from_secs(1);
         const INITIAL_DELAY_MIN_S: u64 = 10;
@@ -431,10 +429,6 @@ You can check your reward balance by running:
 
         let mut high_cpu_count: u8 = 0;
 
-        let process_refresh_kind = ProcessRefreshKind::everything()
-            .without_disk_usage()
-            .without_memory();
-
         // Random initial delay between 1 and 5 minutes
         let initial_delay =
             Duration::from_secs(thread_rng().gen_range(INITIAL_DELAY_MIN_S..=INITIAL_DELAY_MAX_S));
@@ -443,49 +437,37 @@ You can check your reward balance by running:
         loop {
             {
                 let mut system = System::new_all();
-                let physical_core_count = system.physical_core_count();
-                system.refresh_processes_specifics(
-                    ProcessesToUpdate::Some(&[pid]),
-                    true,
-                    process_refresh_kind,
-                );
+                system.refresh_cpu_usage();
                 tokio::time::sleep(Duration::from_millis(300)).await;
-                system.refresh_processes_specifics(
-                    ProcessesToUpdate::Some(&[pid]),
-                    true,
-                    process_refresh_kind,
-                );
-                if let (Some(process), Some(core_count)) =
-                    (system.process(pid), physical_core_count)
-                {
-                    // divide by core_count to get value between 0-100
-                    let cpu_usage = ((process.cpu_usage() as f64 / core_count as f64) * 10000.0)
-                        .round()
-                        / 10000.0;
-                    info!(
-                        "Detected process {pid} CPU usage is {cpu_usage:?} (with {core_count} cores)"
-                    );
-
-                    if cpu_usage > CPU_USAGE_THRESHOLD {
-                        high_cpu_count += 1;
-                    } else {
-                        high_cpu_count = 0;
-                    }
+                system.refresh_cpu_usage();
 
-                    if high_cpu_count >= HIGH_CPU_CONSECUTIVE_LIMIT {
-                        if let Err(err) = ctrl_tx_clone_cpu
-                            .send(NodeCtrl::Stop {
-                                delay: NODE_STOP_DELAY,
-                                result: StopResult::Success(format!("Excess host CPU %{CPU_USAGE_THRESHOLD} detected for {HIGH_CPU_CONSECUTIVE_LIMIT} consecutive minutes!")),
-                            })
-                            .await
-                        {
-                            error!("Failed to send node control msg to antnode bin main thread: {err}");
-                        }
-                        break;
-                    }
+                let cpu_usage = system.global_cpu_usage();
+
+                if cpu_usage > CPU_USAGE_THRESHOLD {
+                    high_cpu_count += 1;
+                } else {
+                    high_cpu_count = 0;
+                }
+
+                info!("Detected Host CPU usage is {cpu_usage:?}.");
+
+                if cpu_usage > CPU_USAGE_THRESHOLD {
+                    high_cpu_count += 1;
                 } else {
-                    error!("Cann't refresh systeminfo of process {pid} with OS core_count of {physical_core_count:?}");
+                    high_cpu_count = 0;
+                }
+
+                if high_cpu_count >= HIGH_CPU_CONSECUTIVE_LIMIT {
+                    if let Err(err) = ctrl_tx_clone_cpu
+                        .send(NodeCtrl::Stop {
+                            delay: NODE_STOP_DELAY,
+                            result: StopResult::Success(format!("Excess host CPU %{CPU_USAGE_THRESHOLD} detected for {HIGH_CPU_CONSECUTIVE_LIMIT} consecutive minutes!")),
+                        })
+                        .await
+                    {
+                        error!("Failed to send node control msg to antnode bin main thread: {err}");
+                    }
+                    break;
                 }
             }
             // Add jitter to the interval