Files
lighthouse/common/health_metrics/src/observe.rs
Mac L a3a74d8988 Correctly compute process times during ProcessHealth::observe (#8793)
I believe I found a bug where during computation of `pid_process_seconds_total` we add `children_system` twice.


  I assume that it was originally intended to add `children_system` and `children_user` once each


Co-Authored-By: Mac L <mjladson@pm.me>
2026-02-14 08:26:25 +00:00

128 lines
4.6 KiB
Rust

use eth2::lighthouse::{Health, ProcessHealth, SystemHealth};
#[cfg(target_os = "linux")]
use {
psutil::cpu::os::linux::CpuTimesExt, psutil::memory::os::linux::VirtualMemoryExt,
psutil::process::Process,
};
pub trait Observe: Sized {
fn observe() -> Result<Self, String>;
}
impl Observe for Health {
#[cfg(not(target_os = "linux"))]
fn observe() -> Result<Self, String> {
Err("Health is only available on Linux".into())
}
#[cfg(target_os = "linux")]
fn observe() -> Result<Self, String> {
Ok(Self {
process: ProcessHealth::observe()?,
system: SystemHealth::observe()?,
})
}
}
impl Observe for SystemHealth {
#[cfg(not(target_os = "linux"))]
fn observe() -> Result<Self, String> {
Err("Health is only available on Linux".into())
}
#[cfg(target_os = "linux")]
fn observe() -> Result<Self, String> {
let vm = psutil::memory::virtual_memory()
.map_err(|e| format!("Unable to get virtual memory: {:?}", e))?;
let loadavg =
psutil::host::loadavg().map_err(|e| format!("Unable to get loadavg: {:?}", e))?;
let cpu =
psutil::cpu::cpu_times().map_err(|e| format!("Unable to get cpu times: {:?}", e))?;
let disk_usage = psutil::disk::disk_usage("/")
.map_err(|e| format!("Unable to disk usage info: {:?}", e))?;
let disk = psutil::disk::DiskIoCountersCollector::default()
.disk_io_counters()
.map_err(|e| format!("Unable to get disk counters: {:?}", e))?;
let net = psutil::network::NetIoCountersCollector::default()
.net_io_counters()
.map_err(|e| format!("Unable to get network io counters: {:?}", e))?;
let boot_time = psutil::host::boot_time()
.map_err(|e| format!("Unable to get system boot time: {:?}", e))?
.duration_since(std::time::UNIX_EPOCH)
.map_err(|e| format!("Boot time is lower than unix epoch: {}", e))?
.as_secs();
Ok(Self {
sys_virt_mem_total: vm.total(),
sys_virt_mem_available: vm.available(),
sys_virt_mem_used: vm.used(),
sys_virt_mem_free: vm.free(),
sys_virt_mem_cached: vm.cached(),
sys_virt_mem_buffers: vm.buffers(),
sys_virt_mem_percent: vm.percent(),
sys_loadavg_1: loadavg.one,
sys_loadavg_5: loadavg.five,
sys_loadavg_15: loadavg.fifteen,
cpu_cores: psutil::cpu::cpu_count_physical(),
cpu_threads: psutil::cpu::cpu_count(),
system_seconds_total: cpu.system().as_secs(),
cpu_time_total: cpu.total().as_secs(),
user_seconds_total: cpu.user().as_secs(),
iowait_seconds_total: cpu.iowait().as_secs(),
idle_seconds_total: cpu.idle().as_secs(),
disk_node_bytes_total: disk_usage.total(),
disk_node_bytes_free: disk_usage.free(),
disk_node_reads_total: disk.read_count(),
disk_node_writes_total: disk.write_count(),
network_node_bytes_total_received: net.bytes_recv(),
network_node_bytes_total_transmit: net.bytes_sent(),
misc_node_boot_ts_seconds: boot_time,
misc_os: std::env::consts::OS.to_string(),
})
}
}
impl Observe for ProcessHealth {
#[cfg(not(target_os = "linux"))]
fn observe() -> Result<Self, String> {
Err("Health is only available on Linux".into())
}
#[cfg(target_os = "linux")]
fn observe() -> Result<Self, String> {
let process =
Process::current().map_err(|e| format!("Unable to get current process: {:?}", e))?;
let process_mem = process
.memory_info()
.map_err(|e| format!("Unable to get process memory info: {:?}", e))?;
let me = procfs::process::Process::myself()
.map_err(|e| format!("Unable to get process: {:?}", e))?;
let stat = me
.stat()
.map_err(|e| format!("Unable to get stat: {:?}", e))?;
let process_times = process
.cpu_times()
.map_err(|e| format!("Unable to get process cpu times : {:?}", e))?;
Ok(Self {
pid: process.pid(),
pid_num_threads: stat.num_threads,
pid_mem_resident_set_size: process_mem.rss(),
pid_mem_virtual_memory_size: process_mem.vms(),
pid_mem_shared_memory_size: process_mem.shared(),
pid_process_seconds_total: process_times.busy().as_secs()
+ process_times.children_system().as_secs()
+ process_times.children_user().as_secs(),
})
}
}