From Bugzilla Helper: User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.8.0.4) Gecko/20060527 SUSE/1.5.0.4-1.9 Firefox/1.5.0.4 Description of problem: When using iostat to monitor performance the %util is seen to go above 100%. The device being monitored is a MegaRAID Logical Volume. Version-Release number of selected component (if applicable): sysstat-5.0.5-14.rhel4 How reproducible: Always Steps to Reproduce: 1. Install an I/O generating package, I used iozone. 2. Launch the I/O generator (./iozone -Ra -s 1G -r 1024 -g 4G -f /home/all1.tmp). 3. Monitor I/O with iostat (iostat -x 1 10 | grep "sda ") Actual Results: [root@l82bi036 ~]# iostat -x 1 10 | grep "sda " Device: rrqm/s wrqm/s r/s w/s rsec/s wsec/s rkB/s wkB/s avgrq-sz avgqu-sz await svctm %util sda 0.05 29.16 0.49 2.59 18.89 254.06 9.44 127.03 88.49 0.89 288.12 0.49 0.15 sda 0.00 125010.89 0.00 2246.53 0.00 1066835.64 0.00 533417.82 474.88 2782.59 195.56 0.44 99.11 sda 0.00 0.00 0.00 2260.00 0.00 0.00 0.00 0.00 0.00 5134.15 993.58 0.44 100.10 sda 0.00 0.00 0.00 2273.00 0.00 0.00 0.00 0.00 0.00 2881.88 1870.48 0.44 100.10 sda 0.00 262.00 0.00 1753.00 0.00 2296.00 0.00 1148.00 1.31 674.11 2602.56 0.46 80.70 sda 0.00 29759.00 0.00 1942.00 0.00 254464.00 0.00 127232.00 131.03 56.42 27.66 0.28 55.10 sda 0.00 35303.00 0.00 2397.00 0.00 301640.00 0.00 150820.00 125.84 108.12 45.27 0.42 100.10 sda 0.00 99921.00 0.00 2271.00 0.00 852704.00 0.00 426352.00 375.48 976.82 98.13 0.44 100.20 sda 0.00 80701.00 0.00 2089.00 0.00 688672.00 0.00 344336.00 329.67 7576.35 763.71 0.48 100.10 sda 0.00 0.00 0.00 2250.00 0.00 0.00 0.00 0.00 0.00 6681.62 1597.33 0.44 100.10 Expected Results: The maximum value for %util should be 100%. Additional info: Server: Dell 6650 with 6 disk internal MegaRAID volume. OS: RHEL4.5 (2.6.9-55.0.12.ELsmp) # df -kh Filesystem Size Used Avail Use% Mounted on /dev/mapper/VolGroup00-LogVol00 165G 12G 145G 8% / /dev/sda1 99M 17M 78M 18% /boot EMC PowerPath is NOT installed on this server.
This problem appears because of in iostat command between the actions "reads data which are used to count the delay" and "read data which are used to count utilization" could be a small time delay which caused this inaccuracy. There is no way how to fix this problem, I discuss this with upstream maintainer and he prefers to leave tis problem as it is. Closing -> wontfix.
*** Bug 590148 has been marked as a duplicate of this bug. ***