Comment 13 for bug 21852

Apologies for yet another post, but I have just tried running iostat to check
what it thinks is happening with the disk subsystem.

I ran this under watch (watch -n 1 iostat), and then put my ear against the HD,
it seeks now and then, but it's not hammering away or anything suspicious,
however, iostat is reporting *alarming* stats:

vaio:~$ iostat
Linux 2.6.12-9-386 (vaio) 28/09/05

avg-cpu: %user %nice %sys %iowait %idle
           6.93 0.11 1.48 16.15 75.33

Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn
sda 41.22 1624.51 78.85 1278410 62049

I monitor iostat on a lot of my Linux servers at work, as they are heavily
utilised DB boxes and application servers. Typical values on another SATA-based
workstation system are around 30-40 reads/second when idle, and perhaps a few
hundred blocks/second when running a clean build of an enterprise Java
application, but never in the region of 1600 reads/second. I think perhaps
iostat might be mis-reporting the figures - but it does make me wonder....
anyone got any thoughts about this??

Thanks,
Mark