iostat: Report Central Processing Unit (CPU) statistics and input/output statistics for devices and partitions
iostat [ -c ] [ -d ] [ -h ] [ -k | -m ] [ -N ] [ -p [ device[,...] | ALL ] ] [ -s ] [ -t ] [ -V ] [ -x ] [ -y ] [ -z ] [ --compact ] [ --dec={ 0 | 1 | 2 } ] [ { -f | +f } directory ] [ -j { ID | LABEL | PATH | UUID | ... } ] [ -o JSON ] [ [ -H ] -g group_name ] [ --human ] [ --pretty ] [ device [...] | ALL ] [ interval [ count ] ]
What it does
iostat reports the CPU summary and per-device I/O. The extended report (-x) is the one for latency: r/s rkB/s rrqm/s %rrqm r_await rareq-sz, the same for writes (w_), discards (d_) and flushes (f_), then aqu-sz and %util.
Options you will use
-x- extended statistics: r_await / w_await (average ms per request, queueing included), aqu-sz (average queue length), %util (share of time the device was busy).
-z- omit devices with no activity in the interval.
-y- omit the first report (the since-boot averages).
-c- the CPU report only.
-d- the device report only.
-p [ device | ALL ]- also the partitions.
-N- device-mapper names (ubuntu--vg-ubuntu--lv) instead of dm-0.
-t- print the time of each report.
-k | -m- kilobytes / megabytes per second.
-h- human-readable sizes (--human) and a pretty layout.
-s- a short (narrow) version of the report.
-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
$ iostat -xz 1extended, busy devices only, every second
$ iostat -xz -N 1with LVM names
$ iostat -c 2 3CPU only
Gotchas
- %util near 100 means saturated only for a device that serves one request at a time. SSDs, NVMe and virtual disks serve many: look at await and aqu-sz.
- The first report is since boot: read the second one (or use -y).
On the blog
Try iostat in a real terminal Free, in your browser - a real Ubuntu terminal to try it in, with missions that check your work.