Expand description
Low-overhead system information collection.
Counter models and parsers work without std or an allocator.
Linux collection is enabled by std; the optional dylib feature adds
shared Linux/macOS/Windows native loading without std or an allocator.
use zsys::cpu::CpuTimes;
let before = CpuTimes { user: 100, idle: 900, ..CpuTimes::default() };
let after = CpuTimes { user: 150, idle: 950, ..CpuTimes::default() };
assert_eq!(after.busy_percent_since(before), Some(50.0));let mut system = zsys::System::new();
system.refresh_memory()?;
system.refresh_processes()?;
for process in system.processes() {
println!("{}: {} bytes", process.id.pid, process.memory_bytes);
}Re-exports§
pub use linux::System;pub use snapshot::Sample;pub use snapshot::SystemSnapshot;
Modules§
- cpu
- Cumulative CPU counters, in kernel clock ticks. Guest time is already in user/nice.
- disk
- Block-device I/O and filesystem-capacity models. No allocation or OS calls.
- linux
- Linux collectors using procfs and native operating-system interfaces.
- memory
- Physical memory and swap counters.
- network
- Allocation-free network counter model and Linux record parsing.
- parse
- Allocation-free procfs parsers. These also compile on non-Linux, no_std targets.
- process
- Process identity and cumulative counters.
- sampling
- Caller-driven scheduling. No background thread, sleep, or hidden refresh.
- snapshot
- Shared snapshot types and sampling metadata.