linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* /proc/partitions statistics question
@ 2003-12-16 19:35 Jean-Luc Fontaine
  2003-12-16 20:37 ` Mike Fedyk
  0 siblings, 1 reply; 2+ messages in thread
From: Jean-Luc Fontaine @ 2003-12-16 19:35 UTC (permalink / raw)
  To: linux-kernel

I have been trying to make sense of the units used in the 
/proc/partitions statistics.
I thought read and write I/O were in kilobytes, but not according to the 
following test on my 2.4.3 kernel machine:

# /proc/partitions columns:
# major minor #blocks name rio rmerge rsect ruse wio wmerge wsect wuse 
running use aveq

$ fgrep hdb1 /proc/partitions
$ dd if=/dev/zero of=ttt bs=1M count=1024 2>/dev/null
$ sync; sync; sync
$ fgrep hdb1 /proc/partitions
    3    65   40209088 hdb1 7107 18180 202278 384530 7925 19005 215616 
215330 0 46240 599860
    3    65   40209088 hdb1 7134 18218 202798 386510 24566 265384 
2319776 2977060 0 102000 3363570
$ ls -lk ttt
-rw-r--r--    1 root     root      1048576 Dec 13 15:56 ttt
$ fgrep hdb1 /proc/partitions
$ dd if=ttt of=/dev/null 2>/dev/null
$ sync; sync; sync
$ fgrep hdb1 /proc/partitions
    3    65   40209088 hdb1 7152 18222 202974 386650 24581 265397 
2320000 2977060 0 102140 3363710
    3    65   40209088 hdb1 23567 263977 2300334 555980 24668 265441 
2321048 2978280 0 171380 3534260

(24566 - 7925) = 16641 were written, whereas
(23567 - 7152) = 16415 were read
I would have expected something near 1048576 (kilobytes),
but the results seem to be roughly 64 times less...

I would really appreciate an explanation, as I use those statistics
in a monitoring program and for filesystems performance tests.

Many thanks for your help!

-- 
Jean-Luc Fontaine  mailto:jfontain@free.fr  http://jfontain.free.fr/

^ permalink raw reply	[flat|nested] 2+ messages in thread

* Re: /proc/partitions statistics question
  2003-12-16 19:35 /proc/partitions statistics question Jean-Luc Fontaine
@ 2003-12-16 20:37 ` Mike Fedyk
  0 siblings, 0 replies; 2+ messages in thread
From: Mike Fedyk @ 2003-12-16 20:37 UTC (permalink / raw)
  To: Jean-Luc Fontaine; +Cc: linux-kernel

On Tue, Dec 16, 2003 at 08:35:06PM +0100, Jean-Luc Fontaine wrote:
> (24566 - 7925) = 16641 were written, whereas
> (23567 - 7152) = 16415 were read
> I would have expected something near 1048576 (kilobytes),
> but the results seem to be roughly 64 times less...
> 
> I would really appreciate an explanation, as I use those statistics
> in a monitoring program and for filesystems performance tests.
> 
> Many thanks for your help!
> 

check out the man page for iostat in the sar/sysstat package.  It documents
what is happening here.   This is in blocks, and there is merging to
consider and etc...

> -- 
> Jean-Luc Fontaine  mailto:jfontain@free.fr  http://jfontain.free.fr/
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/
> 

^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2003-12-16 20:37 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2003-12-16 19:35 /proc/partitions statistics question Jean-Luc Fontaine
2003-12-16 20:37 ` Mike Fedyk

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).