Skip to main content

iostat command man page

IOSTAT(1)      Linux User’s Manual     IOSTAT(1)

NAME
       iostat - Report Central Processing Unit (CPU) statistics and input/out-
       put statistics for devices and partitions.

SYNOPSIS
       iostat [ -c | -d ] [ -k | -m ] [ -t ] [ -V ] [ -x ] [ -n ]  [  -h  ]  [
       device [ ... ] | ALL ] [ -p [ device | ALL ] ] [ interval [ count ] ]

DESCRIPTION
       The  iostat  command  is used for monitoring system input/output device
       loading by observing the time the devices are  active  in  relation  to
       their average transfer rates. The iostat command generates reports that
       can be used to  change  system  configuration  to  better  balance  the
       input/output load between physical disks.

       The  first  report  generated by the iostat command provides statistics
       concerning the time since the system was booted. Each subsequent report
       covers  the time since the previous report. All statistics are reported
       each time the iostat command is run.  The  report  consists  of a  CPU
       header  row followed by a row of CPU statistics. On multiprocessor sys-
       tems, CPU statistics are calculated system-wide as averages  among  all
       processors.  A  device  header  row  is displayed followed by a line of
       statistics for each device that is configured.

       The interval parameter specifies the amount of time in seconds  between
       each  report.  The  first report contains statistics for the time since
       system startup (boot). Each subsequent report contains statistics  col-
       lected during the interval since the previous report. The count parame-
       ter can be specified in conjunction with the interval parameter. If the
       count  parameter is specified, the value of count determines the number
       of reports generated at interval seconds apart. If the interval parame-
       ter is specified without the count parameter, the iostat command gener-
       ates reports continuously.


REPORTS
       The iostat command generates two types of reports, the CPU  Utilization
       report and the Device Utilization report.

       CPU Utilization Report
     The first report generated by the iostat command is the CPU Uti-
     lization Report. For multiprocessor systems, the CPU values  are
     global  averages among all processors.  The report has the fol-
     lowing format:

     %user
    Show the percentage  of  CPU  utilization that  occurred
    while executing at the user level (application).
     %nice
    Show  the percentage  of CPU  utilization that occurred
    while executing at the user level with nice priority.
     %system
    Show the percentage  of  CPU  utilization that  occurred
    while executing at the system level (kernel).
     %iowait
    Show  the percentage  of time that the CPU or CPUs were
    idle during which the system had an outstanding disk  I/O
    request.
     %steal
    Show  the percentage of time spent in involuntary wait by
    the virtual CPU or CPUs while the hypervisor was  servic-
    ing another virtual processor.
     %idle
    Show  the percentage  of time that the CPU or CPUs were
    idle and the system did not have an outstanding disk  I/O
    request.

       Device Utilization Report
     The  second report generated by the iostat command is the Device
     Utilization Report. The device report provides statistics on  a
     per  physical device or partition basis. Block devices for which
     statistics are to be displayed may be  entered  on  the  command
     line. Partitions may also be entered on the command line provid-
     ing that option -x is not used.  If no device nor partition  is
     entered, then statistics are displayed for every device used by
     the system, and providing that the kernel maintains  statistics
     for  it. If the ALL keyword is given on the command line, then
     statistics are displayed for every device defined by the system,
     including those that have never been used.  The report may show
     the following fields, depending on the flags used:

     Device:
    This column gives the device (or partition)  name, which
    is displayed  as hdiskn with  2.2 kernels, for the nth
    device. It is displayed as devm-n with 2.4 kernels, where
    m is the major number of the device, and n a distinctive
    number.  With newer kernels, the device name as listed in
    the /dev directory is displayed.

     tps
    Indicate  the  number  of transfers per second that were
    issued to the device. A transfer is an I/O request to the
    device.  Multiple logical requests can be combined into a
    single I/O request to the device. A transfer is of inde-
    terminate size.

     Blk_read/s
    Indicate the  amount  of data  read  from  the device
    expressed in a number of blocks per  second.  Blocks  are
    equivalent to  sectors  with  2.4 kernels and newer and
    therefore have a size of 512 bytes. With older kernels, a
    block is of indeterminate size.

     Blk_wrtn/s
    Indicate  the  amount  of data  written  to  the device
    expressed in a number of blocks per second.

     Blk_read
    The total number of blocks read.

     Blk_wrtn
    The total number of blocks written.

     kB_read/s
    Indicate  the  amount  of data  read  from  the device
    expressed in kilobytes per second.

     kB_wrtn/s
    Indicate  the  amount  of data  written  to  the device
    expressed in kilobytes per second.

     kB_read
    The total number of kilobytes read.

     kB_wrtn
    The total number of kilobytes written.

     MB_read/s
    Indicate  the  amount  of data  read  from  the device
    expressed in megabytes per second.

     MB_wrtn/s
    Indicate  the  amount  of data  written  to  the device
    expressed in megabytes per second.

     MB_read
    The total number of megabytes read.

     MB_wrtn
    The total number of megabytes written.

     rrqm/s
    The number of read requests merged per second  that  were
    queued to the device.

     wrqm/s
    The  number of write requests merged per second that were
    queued to the device.

     r/s
    The number of read requests  that were  issued  to  the
    device per second.

     w/s
    The  number  of  write  requests  that were issued to the
    device per second.

     rsec/s
    The number of sectors read from the device per second.

     wsec/s
    The number of sectors written to the device per second.

     rkB/s
    The number of kilobytes read from the device per  second.

     wkB/s
    The number of kilobytes written to the device per second.

     rMB/s
    The number of megabytes read from the device per  second.

     wMB/s
    The number of megabytes written to the device per second.

     avgrq-sz
    The average size (in sectors) of the requests  that  were
    issued to the device.

     avgqu-sz
    The average queue length of the requests that were issued
    to the device.

     await
    The average  time (in  milliseconds)  for I/O  requests
    issued to the device to be served. This includes the time
    spent by the requests in queue and the time spent servic-
    ing them.

     svctm
    The  average  service  time  (in  milliseconds)  for  I/O
    requests that were issued to the device.

     %util
    Percentage of CPU time during  which  I/O requests  were
    issued  to the  device  (bandwidth  utilization  for the
    device). Device saturation occurs when  this  value  is
    close to 100%.

     ops/s
    Indicate the number of operations that were issued to the
    mount point per second

     rops/s
    Indicate the number of read operations that  were issued
    to the mount point per second

     wops/s
    Indicate  the number of write operations that were issued
    to the mount point per second


OPTIONS
       -c     The -c option is exclusive of the -d option  and displays  only
     the CPU usage report.

       -d     The  -d  option  is exclusive of the -c option and displays only
     the device utilization report.

       -k     Display statistics in kilobytes per second instead of blocks per
     second. Data  displayed are  valid  only with kernels 2.4 and
     newer.

       -m     Display statistics in megabytes per second instead of blocks  or
     kilobytes per  second.  Data displayed are valid only with ker-
     nels 2.4 and newer.

       -n     Displays the NFS-directory statistic.  Data displayed are valid
     only with kernels 2.6.17 and newer.  This option is exclusive ot
     the -x option.

       -h     Display the NFS report more human readable.

       -p [ { device | ALL } ]
     The -p option is exclusive of the -x option and displays statis-
     tics for block devices and all their partitions that are used by
     the system.  If a device name is entered on  the command line,
     then  statistics for  it and all its partitions are displayed.
     Last, the ALL keyword indicates that statistics have to be  dis-
     played  for  all the block devices and partitions defined by the
     system, including those that have never been  used.   Note  that
     this option works only with post 2.5 kernels.

       -t     Print the time for each report displayed.

       -V     Print version number then exit.

       -x     Display extended statistics.  This option is exclusive of the -p
     and  -n, and  works  with  post 2.5  kernels  since  it needs
     /proc/diskstats  file  or a mounted sysfs to get the statistics.
     This option may also work with older kernels (e.g. 2.4) only  if
     extended statistics are available in /proc/partitions (the ker-
     nel needs to be patched for that).


ENVIRONMENT
       The iostat command takes into account the following  environment vari-
       able:


       S_TIME_FORMAT
     If  this variable  exists and its value is ISO then the current
     locale will be ignored when printing  the date  in  the report
     header.  The  iostat command will use the ISO 8601 format (YYYY-
     MM-DD) instead.


EXAMPLES
       iostat
     Display a single history since  boot  report  for all  CPU  and
     Devices.

       iostat -d 2
     Display a continuous device report at two second intervals.

       iostat -d 2 6
     Display six reports at two second intervals for all devices.

       iostat -x hda hdb 2 6
     Display  six reports of extended statistics at two second inter-
     vals for devices hda and hdb.

       iostat -p sda 2 6
     Display six reports at two second intervals for device  sda  and
     all its partitions (sda1, etc.)

BUGS
       /proc filesystem must be mounted for iostat to work.

       On  SMP machines  iostat assumes that CPU #0 is never disabled. Indeed
       iostat uses it to calculate the time interval.

       Extended statistics are available only with post 2.5 kernels.

FILES
       /proc/stat contains system statistics.

       /proc/partitions contains disk statistics (for  pre  2.5 kernels  that
       have been patched).

       /proc/diskstats contains disks statistics (for post 2.5 kernels).

       /proc/uptime  contains  system  uptime in seconds and centiseconds (for
       post 2.5 kernels).

       /sys contains statistics for block devices (post 2.5 kernels).

AUTHOR
       Sebastien Godard (sysstat <at> wanadoo.fr)

SEE ALSO
       sar(1), mpstat(1), vmstat(8)

       http://perso.orange.fr/sebastien.godard/

Linux OCTOBER 2006     IOSTAT(1)

Comments

Popular posts from this blog