Package: sysstat
Version: 10.0.5-1
Severity: important
`iostat -d -xm 2 300` command sometimes reports the device %util more than
600%, here is some sample data:
Device: rrqm/s wrqm/s r/s w/s rMB/s wMB/s avgrq-sz
avgqu-sz await svctm %util
sda 0.50 1941.50 1672.50 650.50 21.64 12.89 30.44
992.87 419.89 3.04 706.00
sdb 4.00 0.00 4099.00 4.50 52.54 0.03 26.23
221.54 55.95 1.55 637.60
Device: rrqm/s wrqm/s r/s w/s rMB/s wMB/s avgrq-sz
avgqu-sz await svctm %util
sda 0.14 28.47 86.47 12.46 1.09 0.21 27.05
27.92 236.69 1.43 14.16
sdb 0.00 0.57 64.02 0.64 0.93 0.01 29.59
1.10 17.40 2.16 13.97
Device: rrqm/s wrqm/s r/s w/s rMB/s wMB/s avgrq-sz
avgqu-sz await svctm %util
sda 0.00 302.50 478.00 63.00 5.96 0.98 26.26
164.55 351.83 1.85 100.00
sdb 0.00 0.00 50.00 3.50 0.44 0.02 17.79
0.41 7.51 4.15 22.20
Device: rrqm/s wrqm/s r/s w/s rMB/s wMB/s avgrq-sz
avgqu-sz await svctm %util
sda 1.00 1815.00 2733.50 210.50 31.83 7.85 27.60
712.15 236.17 1.32 390.00
sdb 0.00 6.50 1229.00 14.00 15.07 0.09 24.98
18.21 14.65 2.32 288.00
the value of %util 706.00 and 637.60 is so weird, I know the FAQ of sysstat
(http://mcs.une.edu.au/doc/sysstat/FAQ) :
> 3.6. iostat -x displays huge numbers for some fields...
>
> Because of a Linux kernel bug, iostat -x may display huge I/O response times
> (svctm) and a bandwidth utilization (%util) of 100% for some devices. Indeed
> these devices have a value for the field #9 (beginning after the device name)
> in /proc/{partitions,diskstats} which is always different from 0, and even
> negative sometimes. Yet this field should go to zero, since it gives the
> number of I/Os currently in progress (it is incremented as requests are
> submitted, and decremented as they finish).
> To (temporarily) fix the problem, you should reboot your system to reset the
> counters in /proc/{partitions,diskstats}.
Here is the /proc/diskstats
8 0 sda 474944638 2793620 25049714746 548686608 176345975 25328578
65025196237 2921137500 0 1442575668 3491692340
8 1 sda1 2674870 126919 100395082 120325012 245402 327005 7788512
29006272 0 57038148 149376888
8 2 sda2 9 0 56 188 0 0 0 0 0 172 188
8 5 sda5 3836864 482461 34554976 28214880 2862978 4521635 59472496
2138232956 0 63530140 2167976168
8 6 sda6 588700 64237 50975962 84626764 4985183 2429771 106285104
276073212 0 141974932 362510028
8 7 sda7 121279 9470 1951946 2536004 40667276 15024620 568910584
514104360 0 401159104 517504448
8 8 sda8 467722599 2110507 24861833980 312981908 127585136 3025547
64282739541 4258687996 0 1359773060 300705596
8 16 sdb 99373913 93469 2954047151 1172805924 18624102 9319609
3813314137 2412797116 0 263085556 3585802392
8 17 sdb1 5744817 53811 238884970 30411680 11917662 9194145 605638344
265178520 0 73361380 295592352
8 18 sdb2 7 0 38 28 0 0 0 0 0 28 28
8 21 sdb5 93628862 39631 2715160111 1142394088 6706440 125464
3207675793 2147618596 0 197734464 3290269836
But I dont think it's a bug of kernel, because the the value becomes fine after
a moment.
Thanks,
Tanglim LUA
07/21/2014
--
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
