24 comment

1. Please suggest me – system load average is shooting.

   %iowait is high and found with iostat output one disk mentioned below may be the cause.

   Device: rrqm/s wrqm/s r/s w/s rsec/s wsec/s rkB/s wkB/s avgrq-sz avgqu-sz await svctm %util
   cciss/c0d0 2269.60 132.60 296.20 77.60 20521.60 1681.60 10260.80 840.80 59.40 4.15 11.18 2.65 99.20

   please suggest

2. When I enter the command “iostat -d -x 5 3” I do not get any statistics. Here is what I get:

   # iostat -d -x 5 3
   Linux 2.4.33.3 (FILES) 03/01/2008

   Device: rrqm/s wrqm/s r/s w/s rsec/s wsec/s avgrq-sz avgqu-sz await svctm %util

   Device: rrqm/s wrqm/s r/s w/s rsec/s wsec/s avgrq-sz avgqu-sz await svctm %util

   Device: rrqm/s wrqm/s r/s w/s rsec/s wsec/s avgrq-sz avgqu-sz await svctm %util

   Any ideas on what the problem could be? I’m running Slackware 11.0 with the stock kernel.

   Thanks in advance.

   1. where x = device

      example : iostat -x -d sda 5

3. Another good article… thanks for sharing,

4. Mike – you need to not specify “-d”. Only use -d if you’re going to specify a disk like sda or sdb. Just use the -x interval.

   ie, iostat -x 10

   means it will immediately give one output that represents from boot to now, and then every 10 seconds it will output a line that represents the last 10 seconds, and it will repeat forever.

   You can also take this into excel and graph it, paying attention to the 4-5 columns the author mentioned.

5. good artice!!!!
   one question, Please tell what is the use of the below specified files.
   /sys/block/sda/queue/iosched/queued
   /sys/block/sda/queue/nr_requests
   /sys/block/sda/device/queue_depth

6. Hi!

   Thanks for the information, very useful.

   I have one question – what constitutes “high” for r/s & w/s? Is over 10 high? Over 100, 1000?

   I suspect that a hard disk is slowing down on me – whenever there is an I/O intensive operation (swapping, download, starting Firefox or OOo, things like that), everything slows to a crawl, including process switching.

   Is there a way to find out from iostat what bandwidth the disk is giving me? What is the real latency of a disk?

   Thanks!
   Dave.

7. Great article. Thanks!

   I too would like to know what is “high”

   Also, how do you follow on from this. I have had massive IO wait issues at times and tracking down the process causing the issue is very hard. Any tips in this direction?

   Thanks

8. I should use my gravatar email address 🙂

9. Great article. Just a quick note. On one of our older Debian servers we had to install sysstats package to use iostat.

10. Hi,

    My company uses a VPS hosted in softlayer. Anytime Irun iostat, I hadly get any of these data. Could it be a problem with the VPS?
    The sample of the command output for iostat -d -x 5 3 is as below:

    Device:         rrqm/s   wrqm/s   r/s   w/s   rsec/s   wsec/s avgrq-sz avgqu-sz   await  svctm  %util
    
    Device:         rrqm/s   wrqm/s   r/s   w/s   rsec/s   wsec/s avgrq-sz avgqu-sz   await  svctm  %util
    
    Device:         rrqm/s   wrqm/s   r/s   w/s   rsec/s   wsec/s avgrq-sz avgqu-sz   await  svctm  %util
    
    

    Please I will like to know if there is a way around this. Thanks

11. Very useful… thanks!

12. iostat man page :
    The average service time (svctm field) value is meaningless, as I/O statistics are calculated at block level, and we don’t know when the disk driver starts to process a request. For this reason, this field will be removed in a future sysstat version.

13. Please let me know ..how to identify what is causing high disk utilization ??

14. Hey What are the statistics for high – Just information. Need to be more specific.
    If you can not evaluate the information and reach to the conclusion then its of no use.

    Thanks

15. Here’s how you get the information filtered

    iostat -x 5 3 | sed "s: +: :g" | cut -d ' ' -f 1,4,5,13,14  | sed "s: :t:g"
    

    you would have to wait the full 15 seconds before you see any data.
    So this is an even nicer way.

    for n in {1..3}; do iostat -x 1 1 | sed "s: +: :g" | cut -d ' ' -f 1,4,5,13,14  | sed "s: :t:g" | tail -n +2 && sleep 5; done
    

    🙂

16. Hi,

    when i try to capture the disk i/o using iostat in linux server i was getting the below error.
    iostat
    If ‘iostat’ is not a typo you can run the following command to lookup the package that contains the binary:
    command-not-found iostat
    bash: iostat: command not found

    please tel me the alternative command for iostat in Linux.
    i have tried iotop,it was not working.

    Brs,
    Ravi

    1. Debian/Ubuntu: apt-get install sysstat

       CentOS: yum install sysstat

    2. yum install iostat

17. martyn@Atlas ~ $ iotop -d -x 5 10
    Usage: /usr/lib/python-exec/python3.3/iotop [OPTIONS]

    DISK READ and DISK WRITE are the block I/O bandwidth used during the sampling
    period. SWAPIN and IO are the percentages of time the thread spent respectively
    while swapping in and waiting on I/O more generally. PRIO is the I/O priority at
    which the thread is running (set using the ionice command).

    Controls: left and right arrows to change the sorting column, r to invert the
    sorting order, o to toggle the –only option, p to toggle the –processes
    option, a to toggle the –accumulated option, i to change I/O priority, q to
    quit, any other key to force a refresh.

    iotop: error: option -d: invalid floating-point value: ‘-x’

    Take your bad advice and fuck off my internet.

    1. Just use “iotop” , defaults are fine… or read the manual: “man iotop”

18. I use the -d -p options. Check out the man page for iostat as there are several more important options available. I use iostat to keep track of disk writes on my Raspberry Pi to make sure I am not over using my SD card. I have an external USB HDD and have /var and most other folders that get a lot of access there. I run iostat as a cron job every AM and append to a file iostat.txt. This lets me look at the number of writes from day to day and the overall number of writes from boot time. Very helpful for tuning the system.

    Thanks

19. how can I IOSTAT for particular mount like /dev/XXX/YY

20. Could you explain me how to check and optimize Disk latency/transfer times

    Have a question? Post it on our forum!