Linux commands for measuring disk activity

Linux distributions provide quite a few useful commands for gauging disk activity. Here's a look at a few of them.

Linux commands for measuring disk activity
jkfid (CC BY 2.0)
Current Job Listings

Linux systems provide a handy suite of commands for helping you see how busy your disks are, not just how full. In this post, we examine five very useful commands for looking into disk activity. Two of the commands (iostat and ioping) may have to be added to your system, and these same two commands require you to use sudo privileges, but all five commands provide useful ways to view disk activity.

Probably one of the easiest and most obvious of these commands is dstat.

dtstat

In spite of the fact that the dstat command begins with the letter "d", it provides stats on a lot more than just disk activity. If you want to view just disk activity, you can use the -d option. As shown below, you’ll get a continuous list of disk read/write measurements until you stop the display with a ^c. Note that after the first report, each subsequent row in the display will report disk activity in the following time interval, and the default is only one second.

$ dstat -d
-dsk/total-
 read  writ
 949B   73k
  65k     0    <== first second
   0    24k    <== second second
   0    16k
   0	0 ^C

Including a number after the -d option will set the interval to that number of seconds.

$ dstat -d 10
-dsk/total-
 read  writ
 949B   73k
  65k   81M    <== first five seconds
   0    21k    <== second five second
   0  9011B ^C

Notice that the reported data may be shown in a number of different units — e.g., M (megabytes), k (kilobytes), and B (bytes).

Without options, the dstat command is going to show you a lot of other information as well — indicating how the CPU is spending its time, displaying network and paging activity, and reporting on interrupts 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
  0   0 100   0   0| 949B   73k|   0     0 |   0     3B|  38    65
  0   0 100   0   0|   0     0 | 218B  932B|   0     0 |  53    68
  0   1  99   0   0|   0    16k|  64B  468B|   0     0 |  64    81 ^C

The dstat command provides valuable insights into overall Linux system performance, pretty much replacing a collection of older tools, such as vmstat, netstat, iostat, and ifstat, with a flexible and powerful command that combines their features. For more insight into the other information that the dstat command can provide, refer to this post on the dstat command.

iostat

The iostat command helps monitor system input/output device loading by observing the time the devices are active in relation to their average transfer rates. It's sometimes used to evaluate the balance of activity between disks.

$ iostat
Linux 4.18.0-041800-generic (butterfly)         12/26/2018      _x86_64_       (2 CPU)

avg-cpu:  %user   %nice %system %iowait  %steal   %idle
           0.07    0.01    0.03    0.05    0.00   99.85

Device             tps    kB_read/s    kB_wrtn/s    kB_read    kB_wrtn
loop0             0.00         0.00         0.00       1048          0
loop1             0.00         0.00         0.00        365          0
loop2             0.00         0.00         0.00       1056          0
loop3             0.00         0.01         0.00      16169          0
loop4             0.00         0.00         0.00        413          0
loop5             0.00         0.00         0.00       1184          0
loop6             0.00         0.00         0.00       1062          0
loop7             0.00         0.00         0.00       5261          0
sda               1.06         0.89        72.66    2837453  232735080
sdb               0.00         0.02         0.00      48669         40
loop8             0.00         0.00         0.00       1053          0
loop9             0.01         0.01         0.00      18949          0
loop10            0.00         0.00         0.00         56          0
loop11            0.00         0.00         0.00       7090          0
loop12            0.00         0.00         0.00       1160          0
loop13            0.00         0.00         0.00        108          0
loop14            0.00         0.00         0.00       3572          0
loop15            0.01         0.01         0.00      20026          0
loop16            0.00         0.00         0.00         24          0

Of course, all the stats provided on Linux loop devices can clutter the display when you want to focus solely on your disks. The command, however, does provide the -p option, which allows you to just look at your disks — as shown in the commands below.

$ iostat -p sda
Linux 4.18.0-041800-generic (butterfly)         12/26/2018      _x86_64_        (2 CPU)

avg-cpu:  %user   %nice %system %iowait  %steal   %idle
           0.07    0.01    0.03    0.05    0.00   99.85

Device             tps    kB_read/s    kB_wrtn/s    kB_read    kB_wrtn
sda               1.06         0.89        72.54    2843737  232815784
sda1              1.04         0.88        72.54    2821733  232815784

Note that tps refers to transfers per second.

You can also get iostat to provide repeated reports. In the example below, we're getting measurements every five seconds by using the -d option.

$ iostat -p sda -d 5
Linux 4.18.0-041800-generic (butterfly)         12/26/2018      _x86_64_        (2 CPU)

Device             tps    kB_read/s    kB_wrtn/s    kB_read    kB_wrtn
sda               1.06         0.89        72.51    2843749  232834048
sda1              1.04         0.88        72.51    2821745  232834048

Device             tps    kB_read/s    kB_wrtn/s    kB_read    kB_wrtn
sda               0.80         0.00        11.20          0         56
sda1              0.80         0.00        11.20          0         56

If you prefer to omit the first (stats since boot) report, add a -y to your command.

$ iostat -p sda -d 5 -y
Linux 4.18.0-041800-generic (butterfly)         12/26/2018      _x86_64_        (2 CPU)

Device             tps    kB_read/s    kB_wrtn/s    kB_read    kB_wrtn
sda               0.80         0.00        11.20          0         56
sda1              0.80         0.00        11.20          0         56

Next, we look at our second disk drive.

$ iostat -p sdb
Linux 4.18.0-041800-generic (butterfly)         12/26/2018      _x86_64_        (2 CPU)

avg-cpu:  %user   %nice %system %iowait  %steal   %idle
           0.07    0.01    0.03    0.05    0.00   99.85

Device             tps    kB_read/s    kB_wrtn/s    kB_read    kB_wrtn
sdb               0.00         0.02         0.00      48669         40
sdb2              0.00         0.00         0.00       4861         40
sdb1              0.00         0.01         0.00      35344          0

iotop

The iotop command is top-like utility for looking at disk I/O. It gathers I/O usage information provided by the Linux kernel so that you can get an idea which processes are most demanding in terms in disk I/O. In the example below, the loop time has been set to 5 seconds. The display will update itself, overwriting the previous output.

$ sudo iotop -d 5
Total DISK READ:         0.00 B/s | Total DISK WRITE:      1585.31 B/s
Current DISK READ:       0.00 B/s | Current DISK WRITE:      12.39 K/s
  TID  PRIO  USER     DISK READ  DISK WRITE  SWAPIN     IO>    COMMAND
32492 be/4 root        0.00 B/s    0.00 B/s  0.00 %  0.12 % [kworker/u8:1-ev~_power_efficient]
  208 be/3 root        0.00 B/s 1585.31 B/s  0.00 %  0.11 % [jbd2/sda1-8]
    1 be/4 root        0.00 B/s    0.00 B/s  0.00 %  0.00 % init splash
    2 be/4 root        0.00 B/s    0.00 B/s  0.00 %  0.00 % [kthreadd]
    3 be/0 root        0.00 B/s    0.00 B/s  0.00 %  0.00 % [rcu_gp]
    4 be/0 root        0.00 B/s    0.00 B/s  0.00 %  0.00 % [rcu_par_gp]
    8 be/0 root        0.00 B/s    0.00 B/s  0.00 %  0.00 % [mm_percpu_wq]

ioping

The ioping command is an altogether different type of tool, but it can report disk latency — how long it takes a disk to respond to requests — and can be helpful in diagnosing disk problems.

$ sudo ioping /dev/sda1
4 KiB <<< /dev/sda1 (block device 111.8 GiB): request=1 time=960.2 us (warmup)
4 KiB <<< /dev/sda1 (block device 111.8 GiB): request=2 time=841.5 us
4 KiB <<< /dev/sda1 (block device 111.8 GiB): request=3 time=831.0 us
4 KiB <<< /dev/sda1 (block device 111.8 GiB): request=4 time=1.17 ms
^C
--- /dev/sda1 (block device 111.8 GiB) ioping statistics ---
3 requests completed in 2.84 ms, 12 KiB read, 1.05 k iops, 4.12 MiB/s
generated 4 requests in 3.37 s, 16 KiB, 1 iops, 4.75 KiB/s
min/avg/max/mdev = 831.0 us / 947.9 us / 1.17 ms / 158.0 us

atop

The atop command, like top provides a lot of information on system performance, including some stats on disk activity.

ATOP - butterfly      2018/12/26  17:24:19      37d3h13m------ 10ed
PRC | sys    0.03s | user   0.01s | #proc    179 | #zombie    0 | #exit      6 |
CPU | sys       1% | user      0% | irq       0% | idle    199% | wait      0% |
cpu | sys       1% | user      0% | irq       0% | idle     99% | cpu000 w  0% |
CPL | avg1    0.00 | avg5    0.00 | avg15   0.00 | csw      677 | intr     470 |
MEM | tot     5.8G | free  223.4M | cache   4.6G | buff  253.2M | slab  394.4M |
SWP | tot     2.0G | free    2.0G |              | vmcom   1.9G | vmlim   4.9G |
DSK |          sda | busy      0% | read       0 | write      7 | avio 1.14 ms |
NET | transport    | tcpi 4 | tcpo  stall      8 | udpi 1 | udpo 0swout   2255 |
NET | network      | ipi       10 | ipo 7 | ipfrw      0 | deliv      60.67 ms |
NET | enp0s25   0% | pcki      10 | pcko 8 | si    1 Kbps | so    3 Kbp0.73 ms |

  PID SYSCPU  USRCPU  VGROW   RGROW  ST EXC   THR  S CPUNR   CPU  CMD 1/1673e4 |
 3357  0.01s   0.00s   672K    824K  --   -     1  R     0    0%  atop
 3359  0.01s   0.00s     0K      0K  NE   0     0  E     -    0%  <ps>
 3361  0.00s   0.01s     0K      0K  NE   0     0  E     -    0%  <ps>
 3363  0.01s   0.00s     0K      0K  NE   0     0  E     -    0%  <ps>
31357  0.00s   0.00s     0K      0K  --   -     1  S     1    0%  bash
 3364  0.00s   0.00s  8032K    756K  N-   -     1  S     1    0%  sleep
 2931  0.00s   0.00s     0K      0K  --   -     1  I     1    0%  kworker/u8:2-e
 3356  0.00s   0.00s     0K      0K  -E   0     0  E     -    0%  <sleep>
 3360  0.00s   0.00s     0K      0K  NE   0     0  E     -    0%  <sleep>
 3362  0.00s   0.00s     0K      0K  NE   0     0  E     -    0%  <sleep>

If you want to look at just the disk stats, you can easily manage that with a command like this:

$ atop | grep DSK
$ atop | grep DSK
DSK |          sda | busy      0% | read  122901 | write 3318e3 | avio 0.67 ms |
DSK |          sdb | busy      0% | read    1168 | write    103 | avio 0.73 ms |
DSK |          sda | busy      2% | read       0 | write     92 | avio 2.39 ms |
DSK |          sda | busy      2% | read       0 | write     94 | avio 2.47 ms |
DSK |          sda | busy      2% | read       0 | write     99 | avio 2.26 ms |
DSK |          sda | busy      2% | read       0 | write     94 | avio 2.43 ms |
DSK |          sda | busy      2% | read       0 | write     94 | avio 2.43 ms |
DSK |          sda | busy      2% | read       0 | write     92 | avio 2.43 ms |
^C

Being in the know with disk I/O

Linux provides enough commands to give you good insights into how hard your disks are working and help you focus on potential problems or slowdowns. Hopefully, one of these commands will tell you just what you need to know when it's time to question disk performance. Occasional use of these commands will help ensure that especially busy or slow disks will be obvious when you need to check them.

Join the Network World communities on Facebook and LinkedIn to comment on topics that are top of mind.
Related:
Now read: Getting grounded in IoT