#[non_exhaustive]pub struct Pressure {
pub level: PressureLevel,
pub some: Stall,
pub full: Option<Stall>,
pub total: Option<Duration>,
}Expand description
How much work is stalled waiting for a resource (CPU, memory, or I/O).
The best single sign that a machine, or a container, is overloaded: it
measures waiting rather than busyness. A CPU at 100% with nothing stalled is
doing its job; with 30% of the time stalled it is turning work away.
cpu::pressure, mem::pressure,
and disk::pressure return it.
On Linux it is the kernel’s pressure-stall information (PSI, kernel 4.20 and
later), read from the process’s own cgroup when it has one, so a container
sees its own pressure, and from /proc/pressure/ otherwise. Where an OS
reports only a level (macOS and Windows memory), the stall shares are zero
and total is None.
use sys_lib::{self as sys, PressureLevel};
if let Some(p) = sys::cpu::pressure()? {
if p.level >= PressureLevel::Warning {
println!("tasks waited for a CPU {} of the last 10 s", p.some.avg10);
}
}Fields (Non-exhaustive)§
This struct is marked as non-exhaustive
Struct { .. } syntax; cannot be matched against without a wildcard ..; and struct update syntax will not work.level: PressureLevelThe level the stall shares amount to (see the thresholds).
some: StallThe share of time at least one task was stalled.
full: Option<Stall>The share of time every non-idle task was stalled at once (None for
CPU at the host level, where the kernel does not report it).
total: Option<Duration>The cumulative time tasks were stalled (some), when reported.
Implementations§
Source§impl Pressure
impl Pressure
Sourcepub const WARNING_SOME_AVG10: f32 = 10.0
pub const WARNING_SOME_AVG10: f32 = 10.0
some.avg10 at or above this is PressureLevel::Warning.
Sourcepub const CRITICAL_SOME_AVG10: f32 = 40.0
pub const CRITICAL_SOME_AVG10: f32 = 40.0
some.avg10 at or above this is PressureLevel::Critical.
Sourcepub const CRITICAL_FULL_AVG10: f32 = 10.0
pub const CRITICAL_FULL_AVG10: f32 = 10.0
full.avg10 above this is PressureLevel::Critical.