mpstat: Report processors related statistics
mpstat [ -A ] [ -n ] [ -T ] [ -u ] [ -I { keyword[,...] | ALL } ] [ -N { node_list | ALL } ] [ -o JSON ] [ -P { cpu_list | ALL } ] [ interval [ count ] ]
What it does
mpstat writes the time spent by the CPUs in each mode (%usr %nice %sys %iowait %irq %soft %steal %guest %gnice %idle), for all CPUs together or per CPU with -P. With an interval it prints one report per interval and an Average line at the end (or at Ctrl+C).
Options you will use
-P { cpu_list | ALL }- per-processor statistics: ALL = every CPU plus the "all" line. The way to see one CPU at 100% while the rest idle.
-u- CPU utilization (the default report).
-I { SUM | CPU | SCPU | ALL }- interrupt statistics.
-n- summary CPU statistics by NUMA node.
-T- topology elements (socket and core) in the CPU report.
-A- same as -n -u -I ALL -N ALL -P ALL.
-o JSON- JSON output.
-V- print version number then exit.
interval [count]- report every INTERVAL seconds, COUNT times (no count = until Ctrl+C). The first report covers the time since boot (iostat: unless -y).
Examples
$ mpstat -P ALL 1every CPU, every second
$ mpstat 2 5all CPUs together, five 2-second reports
Gotchas
- %iowait is idle time with I/O outstanding, not CPU work.
- %steal is time a virtual CPU waited for the hypervisor (another guest ran).
- Colours (S_COLORS=auto on a terminal): high values bold red/magenta, zeros dim blue.
On the blog
Try mpstat in a real terminal Free, in your browser - a real Ubuntu terminal to try it in, with missions that check your work.