sysinfo

sysinfo is a crate used to get a system's information on Linux.

Supported OSes

This crate supports Linux only. It has been refactored to focus exclusively on Linux systems, providing comprehensive system information including:

• Process information
• CPU usage and details
• Memory and swap information
• Disk information
• Network statistics
• Hardware components (temperature sensors, etc.)
• GPU information

The minimum-supported version of rustc is 1.95.

Usage

If you want to migrate from an older version, don't hesitate to take a look at the CHANGELOG and at the migration guide.

⚠️ Before any attempt to read the different structs' information, you need to update them to get up-to-date information because for most of them, it works on diff between the current value and the old one.

Which is why, it's much better to keep the same instance of [System] around instead of recreating it multiple times.

You have an example into the examples folder. You can run it with cargo run --example simple.

Otherwise, here is a little code sample:

use sysinfo::{
    Components, Disks, Networks, System,
};

// Please note that we use "new_all" to ensure that all lists of
// CPUs and processes are filled!
println!("=> system:");

// Display system information:
println!("System name:             {:?}", System::name());
println!("System kernel version:   {:?}", System::kernel_version());
println!("System OS version:       {:?}", System::os_version());
println!("System host name:        {:?}", System::host_name());

if let Ok(mut sys) = System::new_all() {
    // First we update all information of our `System` struct.
    sys.refresh_all();

    // RAM and swap information:
    println!("total memory: {} bytes", sys.total_memory());
    println!("used memory : {} bytes", sys.used_memory());
    println!("total swap  : {} bytes", sys.total_swap());
    println!("used swap   : {} bytes", sys.used_swap());

    // Number of CPUs:
    println!("NB CPUs: {}", sys.cpus().len());

    // Display processes ID, name and disk usage:
    for (pid, process) in sys.processes() {
        println!("[{pid}] {:?} {:?}", process.name(), process.disk_usage());
    }
}

// We display all disks' information:
println!("=> disks:");
if let Ok(disks) = Disks::new_with_refreshed_list() {
    for disk in &disks {
        println!("{disk:?}");
    }
}

// Network interfaces name, total data received and total data transmitted:
if let Ok(networks) = Networks::new_with_refreshed_list() {
    println!("=> networks:");
    for (interface_name, data) in &networks {
        println!(
            "{interface_name}: {} B (down) / {} B (up)",
            data.total_received(),
            data.total_transmitted(),
        );
        // If you want the amount of data received/transmitted since last call
        // to `Networks::refresh`, use `received`/`transmitted`.
    }
}

// Components temperature:
if let Ok(components) = Components::new_with_refreshed_list() {
    println!("=> components:");
    for component in &components {
        println!("{component:?}");
    }
}

Please remember that to have some up-to-date information, you need to call the equivalent refresh method. For example, for the CPU usage:

use sysinfo::System;

let Ok(mut sys) = System::new() else { return };

loop {
    sys.refresh_cpu_usage(); // Refreshing CPU usage.
    for cpu in sys.cpus() {
        print!("{}% ", cpu.usage());
    }
    // Sleeping to let time for the system to run for long
    // enough to have useful information.
    std::thread::sleep(sysinfo::MINIMUM_CPU_UPDATE_INTERVAL);
}

With the multithread feature (which is not enabled by default), sysinfo uses multiple threads. However, this can increase the memory usage on some platforms (macOS for example).

Good practice / Performance tips

Most of the time, you don't want all information provided by sysinfo but just a subset of it. In this case, it's recommended to use refresh_specifics(...) methods with only what you need to have much better performance.

Another issue frequently encountered: unless you know what you're doing, it's almost all the time better to instantiate the System struct once and use this one instance through your program. The reason is because a lot of information needs a previous measure to be computed (the CPU usage for example). Another example why it's much better: in case you want to list all running processes, sysinfo needs to allocate all memory for the Process struct list, which takes quite some time on the first run.

If your program needs to use a lot of file descriptors, you'd better use:

sysinfo::set_open_files_limit(0);

as sysinfo keeps a number of file descriptors open to have better performance on some targets when refreshing processes.

Running on Raspberry Pi

The crate works on Raspberry Pi as it runs Linux. If you need to cross-build for ARM, you can follow these steps:

First install the arm toolchain, for example on Ubuntu:

> sudo apt-get install gcc-multilib-arm-linux-gnueabihf

Then configure cargo to use the corresponding toolchain:

cat << EOF > ~/.cargo/config
[target.armv7-unknown-linux-gnueabihf]
linker = "arm-linux-gnueabihf-gcc"
EOF

Finally, cross compile:

rustup target add armv7-unknown-linux-gnueabihf
cargo build --target=armv7-unknown-linux-gnueabihf

Linux on Docker & Windows Subsystem for Linux (WSL)

Virtual Linux systems, such as those run through Docker and Windows Subsystem for Linux (WSL), do not receive host hardware information via /sys/class/hwmon or /sys/class/thermal. As such, querying for components may return no results (or unexpected results) when using this library on virtual systems.

How it works

I wrote a blog post you can find here which explains how sysinfo extracts information on the different systems.

Running tests

Because we're looking at system information, some tests have a better chance to succeed when there is a limited number of parallel running tests. To ensure they all pass, use:

cargo test -- --test-threads=1

C interface

It's possible to use this crate directly from C. Take a look at the Makefile and at the examples/simple.c file.

To build the C example, just run:

> make
> ./simple
# If needed:
> LD_LIBRARY_PATH=target/debug/ ./simple

Benchmarks

You can run the benchmarks locally with rust nightly by doing:

> cargo bench

Donations

If you appreciate my work and want to support me, you can do it with github sponsors or with patreon.