17.12.2017

The Advent of Void: Day 17: dstat

There are many ways to see what’s going on your system, such as top(1), htop(1), atop(1). However, these tools only show the current state of the system, and it’s not so easy to see how the data changes.

dstat(1) works differently: it prints a line every second with some system stats you can configure. By default, it shows CPU usage, disk I/O, network traffic, paging, and context switches:

% dstat
You did not select any stats, using -cdngy by default.
--total-cpu-usage-- -dsk/total- -net/total- ---paging-- ---system--
usr sys idl wai stl| read  writ| recv  send|  in   out | int   csw 
 45   9  45   1   0| 284k  181k|   0     0 |  18k   36k| 754  2181 
 28   6  65   1   0|   0     0 |4698k  222k|   0     0 |4123    17k
 34   8  58   0   0|   0     0 |4730k  220k|   0     0 |5253    20k
 29   8  62   1   0|4096B 1080k|4640k  215k|   0     0 |4383    18k
 37   7  38  17   0|   0    15M|4844k  219k|   0     0 |5532    21k
 31   8  30  32   0| 336k 6840k|5153k  232k|   0     0 |4383    17k
...

For Void development, I like to run this configuration; showing per-core CPU usage, memory usage, process spawn rate and I/O.

% dstat --cpu-use -mpr
---------per-cpu-usage--------- ------memory-usage----- ---procs--- --io/total-
 0   1   2   3   4   5   6   7 | used  free  buff  cach|run blk new| read  writ
  4   2   1   1   1   1   1   1| 550M 2510M 2823M 1315M|  0   0 4.4|6.42  0.93 
  1   0   0   0   0   1   0   0| 550M 2510M 2823M 1315M|  0   0 1.0|   0     0 
  2   2   1   1   0   0   0   1| 551M 2508M 2823M 1315M|  0   0  20|   0     0 
  1   0   0   0   0   0   0   0| 551M 2508M 2823M 1315M|  0   0   0|   0     0 
  0   0   0   0   0   0   0   0| 551M 2509M 2823M 1315M|  0   0   0|   0     0 
  2   0   0   0   0   1   0   0| 551M 2509M 2823M 1315M|  0   0   0|   0     0 

This allows us to easily detect if builds are not optimally parallelized, or whether we run into memory limits. And process spawn rate detects if a ./configure script runs. ;)

dstat includes a huge list of probes it can display, just to list a few: battery usage, free disk, MySQL/InnoDB stats, qmail queue size, NFS traffic, NTP stats, WiFi measurements, ZFS information.

It’s a very versatile tool, and because it works via the console, you can also easily run it over SSH.