Linux User ' s Manual IOSTAT

<span id="Label3"></p><p><p>IOSTAT (1) Linux User ' s Manual IOSTAT (1)</p></p><p><p>NAME<br>Iostat-report Central processing Unit (CPU) statistics and input/out-<br>Put statistics for devices, partitions and network filesystems (NFS).</p></p><p><p>Synopsis<br>Iostat [-c] [-d] [-n] [-n] [-h] [-k |-m] [-t] [-v] [<br>-x] [-z] [device [...] | all] [-p [device [,...] | all]] [<br>Interval [count]]</p></p><p><p>DESCRIPTION<br>The Iostat command is used for monitoring system input/output device<br>Loading by observing the time the devices is active in relation to<br>Their average transfer rates. The Iostat command generates reports that<br>Can is used to change system configuration to better balance the<br>Input/output load between physical disks.</p></p><p><p> The first report generated by the Iostat command provides statistics <br> concerning the time since the system was boot Ed. Each subsequent report <br> covers the time since the previous Report. All statistics be reported <br> each time the Iostat command is Run. The report consists of a CPU <br> header row followed by a row of CPU Statistics. On multiprocessor sys-<br> tems, CPU statistics is calculated system-wide as averages among all <br> Processors. A device header row is displayed followed by a line of <br> statistics for each device, which is Configured. When Option-n was used, <br> An NFS header row was displayed followed by a line of statistics for <br> each network filesyste M is Mounted. </p></p><p><p>The interval parameter specifies the amount of time in seconds between<br>Each report. The first report contains statistics for the time since<br>System Startup (boot). Each subsequent report contains statistics col-<br>lected during the interval since the previous Report. The Count parame-<br>TER can is specified in conjunction with the interval parameter. If the<br>Count parameter is specified, the value of the count determines the number<br>of reports generated at interval seconds apart. If the interval parame-<br>TER is specified without the Count parameter, the Iostat command gener-<br>Ates Reports Continuously.</p></p><p><p>REPORTS<br>The Iostat command generates three types of reports, the CPU utiliza-<br>tion report, The Device Utilization report and the Network Filesystem<br>Report.</p></p><p><p>CPU Utilization Report<br>The first report generated by the Iostat command is the CPU uti-<br>Lization Report. For multiprocessor systems, the CPU values is<br>Global averages among all Processors. The report has the fol-<br>lowing format:</p></p><p><p>%user<br>Show the percentage of CPU utilization that occurred<br>While executing at the user level (application).</p></p><p><p>%nice<br>Show the percentage of CPU utilization that occurred<br>While executing is at the user level with a nice Priority.</p></p><p><p>%system<br>Show the percentage of CPU utilization that occurred<br>While executing at the system level (kernel).</p></p><p><p>%iowait<br>Show the percentage of time that the CPU or CPUs were<br>Idle during which the system had an outstanding disk I/O<br>Request.</p></p><p><p>%steal<br>Show the percentage of time spent in involuntary wait by<br>The virtual CPU or CPUs while the hypervisor is servic-<br>ing another virtual Processor.</p></p><p><p>%idle<br>Show the percentage of time that the CPU or CPUs were<br>Idle and the system did not has an outstanding disk I/O<br>Request.</p></p><p><p> Device Utilization Report <br> The second report generated by the Iostat command was the device <br> Utilization Report. The device report provides statistics on a <br> per physical device or partition Basis. Block devices for which <br> statistics is displayed may is entered on the command <br> Line. Partitions may also was entered on the command line provid-<br> ing that option-x was not used. IF no device nor partition is <br> entered, then statistics be displayed for every device used by <br> the system, and Prov IDing that the kernel maintains statistics <br> for it. If The All keyword was given on the command line and then <br> Statistics was displayed for every device defined by the system, <br> including those that has never been used. The report could show <br> the following fields, depending the flags used: </p></p><p><p>Device:<br>This column gives the device (or Partition) name, which<br>is displayed as HDISKN with 2.2 kernels, for the nth<br>Device. It is displayed as Devm-n with 2.4 kernels, where<br>M is the major number of the device, and n a distinctive<br>Number. With newer kernels, the device name as listed in<br>The/dev Directory is Displayed.</p></p><p><p>Tps<br>Indicate the number of transfers per second that were<br>Issued to the Device. A transfer is an I/O request to the<br>Device. Multiple logical requests can be combined into a<br>Single I/O request to the DEVICE. A transfer is of inde-<br>Terminate Size.</p></p><p><p>blk_read/s<br>Indicate the amount of data read from the device<br>Expressed in a number of blocks per Second. Blocks is<br>Equivalent to sectors with kernels 2.4 and later and<br>Therefore has a size of bytes. With older kernels, a<br>Block is of indeterminate size.</p></p><p><p>blk_wrtn/s<br>Indicate the amount of data written to the device<br>Expressed in a number of blocks per Second.</p></p><p><p>Blk_read<br>The total number of blocks Read.</p></p><p><p>Blk_wrtn<br>The total number of blocks Written.</p></p><p><p>kb_read/s<br>Indicate the amount of data read from the device<br>Expressed in kilobytes per Second.</p></p><p><p>kb_wrtn/s<br>Indicate the amount of data written to the device<br>Expressed in kilobytes per Second.</p></p><p><p>Kb_read<br>The total number of kilobytes Read.</p></p><p><p>Kb_wrtn<br>The total number of kilobytes Written.</p></p><p><p>mb_read/s<br>Indicate the amount of data read from the device<br>expressed in megabytes per Second.</p></p><p><p>mb_wrtn/s<br>Indicate the amount of data written to the device<br>expressed in megabytes per Second.</p></p><p><p>Mb_read<br>The total number of megabytes Read.</p></p><p><p>Mb_wrtn<br>The total number of megabytes Written.</p></p><p><p>rrqm/s<br>The number of read requests merged per second that were<br>Queued to the Device.</p></p><p><p>wrqm/s<br>The number of write requests merged per second that were<br>Queued to the Device.</p></p><p><p>r/s<br>The number of read requests that were issued to the<br>Device per Second.</p></p><p><p>w/s<br>The number of write requests that were issued to the<br>Device per Second.</p></p><p><p>rsec/s<br>The number of sectors read from the device per second.</p></p><p><p>wsec/s<br>The number of sectors written to the device per Second.</p></p><p><p>rkb/s<br>The number of kilobytes read from the device per second.</p></p><p><p>wkb/s<br>The number of kilobytes written to the device per Second.</p></p><p><p>rmb/s<br>The number of megabytes read from the device per second.</p></p><p><p>wmb/s<br>The number of megabytes written to the device per Second.</p></p><p><p>Avgrq-sz<br>The average size (in sectors) of the requests that were<br>Issued to the Device.</p></p><p><p>Avgqu-sz<br>The average queue length of the requests that were issued<br>to the Device.</p></p><p><p>Await<br>The average time (in Milliseconds) for I/O requests<br>Issued to the device to be Served. This includes the time<br>Spent by the requests in queue and the time spent servic-<br>ing Them.</p></p><p><p>Svctm<br>The average service time (in Milliseconds) for I/O<br>Requests that were issued to the DEVICE. warning! Do not<br>Trust this field any More. This field would be is removed in<br>A future Sysstat Version.</p></p><p><p>%util<br>Percentage of CPU time during which I/O requests were<br>Issued to the device (bandwidth utilization for the<br>Device). Device saturation occurs when the this value is<br>Close to 100%.</p></p><p><p>Network Filesystem Report<br>The Network Filesystem (NFS) report provides statistics for each<br>Mounted Network Filesystem. The report shows the following<br>Fields</p></p><p><p>Filesystem:<br>This columns shows the hostname of the NFS server fol-<br>Lowed by a colon and by the directory name where the net-<br>Work filesystem is Mounted.</p></p><p><p>rblk_nor/s<br>Indicate the number of blocks read by applications via<br>The read (2) system call Interface. A block has a size of<br>Bytes.</p></p><p><p>wblk_nor/s<br>Indicate the number of blocks written by applications via<br>The Write (2) system call Interface.</p></p><p><p>rblk_dir/s<br>Indicate the number of blocks read from files opened with<br>The O_direct Flag.</p></p><p><p>wblk_dir/s<br>Indicate the number of blocks written to files opened<br>With the O_direct Flag.</p></p><p><p>rblk_svr/s<br>Indicate the number of blocks read from the server by the<br>NFS Client via an NFS READ Request.</p></p><p><p>wblk_svr/s<br>Indicate the number of blocks written to the server by<br>The NFS client via an NFS WRITE Request.</p></p><p><p>rkb_nor/s<br>Indicate the number of kilobytes read by applications via<br>The read (2) system call Interface.</p></p><p><p>wkb_nor/s<br>Indicate the number of kilobytes written by applications<br>Via the Write (2) system call Interface.</p></p><p><p>rkb_dir/s<br>Indicate the number of kilobytes read from files opened<br>With the O_direct Flag.</p></p><p><p>wkb_dir/s<br>Indicate the number of kilobytes written to files opened<br>With the O_direct Flag.</p></p><p><p>rkb_svr/s<br>Indicate the number of kilobytes read from the server by<br>The NFS client via an NFS READ Request.</p></p><p><p>wkb_svr/s<br>Indicate the number of kilobytes written to the server by<br>The NFS client via an NFS WRITE Request.</p></p><p><p>rmb_nor/s<br>Indicate the number of megabytes read by applications via<br>The read (2) system call Interface.</p></p><p><p>wmb_nor/s<br>Indicate the number of megabytes written by applications<br>Via the Write (2) system call Interface.</p></p><p><p>rmb_dir/s<br>Indicate the number of megabytes read from files opened<br>With the O_direct Flag.</p></p><p><p>wmb_dir/s<br>Indicate the number of megabytes written to files opened<br>With the O_direct Flag.</p></p><p><p>rmb_svr/s<br>Indicate the number of megabytes read from the server by<br>The NFS client via an NFS READ Request.</p></p><p><p>wmb_svr/s<br>Indicate the number of megabytes written to the server by<br>The NFS client via an NFS WRITE Request.</p></p><p><p>ops/s<br>Indicate the number of operations that were issued to the<br>FileSystem per Second.</p></p><p><p>rops/s<br>Indicate the number of ' read ' operations that were issued<br>To the filesystem per Second.</p></p><p><p>wops/s<br>Indicate the number of ' write ' operations that were<br>Issued to the filesystem per Second.</p></p><p><p>OPTIONS<br>-c Display the CPU utilization Report.</p></p><p><p>-d Display the device utilization Report.</p></p><p><p>-h make the NFS report displayed by Option-n easier to read by a<br>Human.</p></p><p><p>-k Display Statistics in kilobytes per second instead of blocks per<br>Second. Data displayed is valid only with kernels 2.4 and<br>Later.</p></p><p><p>-m Display Statistics in megabytes per second instead of blocks or<br>Kilobytes per Second. Data displayed is valid only with ker-<br>Nels 2.4 and Later.</p></p><p><p>-n Display The registered device mapper names for any device mapper<br>Devices. Useful for viewing LVM2 statistics.</p></p><p><p>-n Display The network filesystem (NFS) Report. This option works<br>Only with kernel 2.6.17 and later.</p></p><p><p>-p [{device [,...] | all}]<br>THE-P option displays statistics for block devices and all<br>Their partitions that is used by the System. If a device name<br>is entered on the command line, then statistics for it and all<br>Its partitions is Displayed. last, The ALL keyword indicates<br>That statistics has the displayed for all the block devices<br>and partitions defined by the system, including Those.<br>Never been Used. Note that this option works on with post 2.5<br>Kernels.</p></p><p><p>-t Print the time for each report Displayed. The timestamp format<br>May depend on the value of the S_time_format environment vari-<br>Able (see below).</p></p><p><p>-v Print version number then exit.</p></p><p><p>-x Display Extended Statistics. This option works with post 2.5<br>Kernels since it needs/proc/diskstats file or a mounted SYSFS<br>To get the Statistics. This option is also work with older ker-<br>Nels (E.G. 2.4) only if extended statistics is available in<br>/proc/partitions (the kernel needs to is patched for).</p></p><p><p>-z tell Iostat to omit output for any devices for which there was<br>No activity during the sample Period.</p></p><p><p>Environment<br>The Iostat command takes into account the following environment vari-<br>Able</p></p><p><p>S_time_format<br>If this variable exists and its value are ISO then the current<br>Locale is ignored when printing the date<br>Header. The Iostat command would use the ISO 8601 format (yyyy-<br>Mm-dd) Instead. The timestamp displayed with OPTION-T would<br>Also is compliant with ISO 8601 format.</p></p><p><p>EXAMPLES<br>Iostat<br>Display a single history since boot report for all CPUs and<br>Devices.</p></p><p><p>Iostat-d 2<br>Display A continuous device report at the second Intervals.</p></p><p><p>Iostat-d 2 6<br>Display six reports at the second intervals for all Devices.</p></p><p><p>Iostat-x HDA HDB 2 6<br>Display Six reports of extended statistics at the second inter-<br>Vals for Devices Hda and Hdb.</p></p><p><p>Iostat-p SDA 2 6<br>Display six reports at the second intervals for device SDA and<br>All its partitions (sda1, etc.)</p></p><p><p>BUGS<br>/proc filesystem must is mounted for Iostat to Work.</p></p><p><p>Extended statistics is available only with post 2.5 kernels.</p></p><p><p>The average service time (SVCTM Field) value is meaningless, as I/O<br>Statistics is calculated at Block level, and we don ' t know when the<br>Disk driver starts to process a Request. For this reason, the This field<br>'ll is removed in a future Sysstat Version.</p></p><p><p>FILES<br>/proc/stat contains system Statistics.</p></p><p><p>/proc/uptime contains system Uptime.</p></p><p><p>/proc/partitions contains disk statistics (for pre 2.5 kernels that<br>has been patched).</p></p><p><p>/proc/diskstats contains disks statistics (for post 2.5 kernels).</p></p><p><p>/sys contains statistics for block devices (post 2.5 kernels).</p></p><p><p>/proc/self/mountstats contains statistics for network Filesystems.</p></p><p><p>AUTHOR<br>Sebastien Godard (sysstat <at> Orange.fr)</p></p><p><p>See ALSO<br>SAR (1), pidstat (1), mpstat (1), vmstat (8)</p></p><p><p>http://pagesperso-orange.fr/sebastien.godard/</p></p><p><p>Linux APRIL IOSTAT (1)</p></p><p><p>Linux User ' s Manual IOSTAT</p></p></span>