Canonical new pressure metrics for three major resources—memory, CPU, and IO.

Detect resource pressure increases in advance and take proactive steps like pausing or killing tasks, or reallocating memory.

See pressure metrics on a per-cgroup basis and respond intelligently, before system crashes.