CLOCKMONOTONIC

CLOCK_MONOTONIC is a clock ID used with POSIX time APIs such as clock_gettime, clock_getres, and clock_nanosleep to obtain a clock that advances monotonically. It represents time since an unspecified starting point, typically the system boot, and is not tied to the wall clock time.

A key characteristic is that CLOCK_MONOTONIC is not affected by changes to the system time or time-of-day corrections. It will not go backward if the user or network adjusts the clock, making it suitable for measuring elapsed time and implementing timeouts. It is designed to advance at a steady rate, suitable for duration calculations rather than real-world timestamps.

On many systems, CLOCK_MONOTONIC does not advance while the machine is suspended; time resumes from the moment of wake-up. Related clocks exist for different use cases, such as CLOCK_MONOTONIC_RAW, which is not adjusted by the kernel’s time discipline (for raw hardware timing), and CLOCK_BOOTTIME, which includes time spent in suspend.

Usage considerations include that CLOCK_MONOTONIC is not appropriate for obtaining real-world timestamps or for scheduling events against an actual date. For elapsed-time measurements, applications typically read two timestamps with CLOCK_MONOTONIC and compute their difference. The returned value is a timespec with seconds and nanoseconds, enabling high-resolution interval calculations.

Availability and standards: CLOCK_MONOTONIC is defined by POSIX and is widely implemented on Unix-like systems, including Linux, BSD variants, and macOS. While generally reliable for interval timing, its exact behavior during edge cases (such as sleep states and hardware timer quirks) can vary by platform, so system-specific documentation may provide additional details.