Re: Eradic disk access during reads

[Denis Vlasenko <vda@ilport.com.ua>]

[-- Attachment #1: Type: text/plain, Size: 1315 bytes --]

> > > Monitoring disk access using gkrellm, I noticed that a command like
> > >
> > > cat /dev/hda > /dev/null
> > >
> > > shows eradic disk reads ranging from 0 to 80MB/s on an otherwise idle
> > > system.
> > >
> > > 1. Is this a hardware or software problem?
> >
> > Difficult to tell without more info. Can be a broken IDE disk or defective
> > ribbon.
> 
> Tried the same with 2.4.31 which shows steady behaviour with occasional dips 
> and pops in the msec range.
> 
> > > 2. Is there a lightweight perf-mon tool (cmd-line) that would log this
> > > behaviour graphically?
> >
> > You can do " readspeed </dev/hda | tr '\r' '\n' > log " with the readspeed
> > tool from there :
> >    http://w.ods.org/tools/readspeed

Try attached one.

# dd if=/dev/hda of=/dev/null bs=16M

On another console:

# nmeter t6 b d1
...
13:42:28.352037 bio    0    0
13:42:28.354675 bio    0    0
13:42:28.357703 bio    0    0
13:42:28.360671 bio    0    0
13:42:28.364712 bio    0    0
13:42:28.367671 bio 128k    0
13:42:28.370673 bio    0    0
13:42:28.373656 bio    0    0
13:42:28.376653 bio    0    0
13:42:28.379680 bio 128k    0
13:42:28.382656 bio    0    0
13:42:28.385819 bio    0    0
13:42:28.388651 bio    0    0
13:42:28.391736 bio 128k    0
13:42:28.394673 bio    0    0
13:42:28.397656 bio    0    0
...
--
vda

[-- Attachment #2: d.tar.bz2 --]
[-- Type: application/x-tbz, Size: 19271 bytes --]