Struct Telemetry 

pub struct Telemetry {
    pub timestamp: u64,
    pub system: SystemInfo,
    pub hardware: HardwareInfo,
    pub network: NetworkInfo,
}

Full system telemetry snapshot.

Contains three sub-structures: SystemInfo, HardwareInfo, and NetworkInfo, plus a Unix timestamp. Service detection has been removed in favor of raw listening ports in NetworkInfo.

Fields

timestamp: u64

Unix timestamp (seconds) when the snapshot was taken.

system: SystemInfo

Basic system information (CPU, RAM, disk, uptime, load).

hardware: HardwareInfo

Special hardware devices (TPU, GPU, JAX availability).

network: NetworkInfo

Network state (public IP, tunnel status, listening ports).

Implementations

impl Telemetry

pub fn capture() -> Self

Captures a full system telemetry snapshot.

Results are cached for 30 seconds to avoid repeated filesystem reads on consecutive calls. Network queries (public_ip, tunnel) are included in the cached value.

Use capture_lightweight for execution paths that don’t need accelerator detection or network probing (e.g., the executor’s WAL audit trail — which only needs /proc-based system health data).

pub fn capture_lightweight() -> Self

Captures a lightweight system telemetry snapshot without shell-outs.

Unlike capture, this method skips all accelerator detection (rocm-smi, nvidia-smi, JAX import), tunnel probing, and public IP queries. Only SystemInfo is populated from /proc reads and a single df shell-out.

HardwareInfo is zeroed (empty accelerators, no JAX), and NetworkInfo returns defaults (public_ip = "unknown", tunnel_running = false, empty listening_ports).

Use this in hot paths like the executor’s WAL audit trail where GPU/TPU/JAX counts are irrelevant and shell-outs produce unwanted stderr noise on systems without those tools installed.

Results share the same internal cache — a previous full capture within 30 seconds will be returned AS-IS (including hardware/network data). Callers on hot paths should NOT rely on this returning empty hardware if a full capture was recently cached.

pub fn clear_cache()

Clears the telemetry cache.

Call this in tests or between full/lightweight captures to prevent stale cached data from leaking across capture modes.

pub fn print_report(&self)

Prints telemetry in a human-readable report to stdout.

Output includes CPU cores, RAM available, machine-parseable uptime seconds, contextualized load average (with core count), raw listening ports, and tunnel PID.

Trait Implementations

impl Clone for Telemetry

fn clone(&self) -> Telemetry

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for Telemetry

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl<'de> Deserialize<'de> for Telemetry

fn deserialize<__D>(__deserializer: __D) -> Result<Self, __D::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Serialize for Telemetry

fn serialize<__S>(&self, __serializer: __S) -> Result<__S::Ok, __S::Error>

where __S: Serializer,

Serialize this value into the given Serde serializer.