Flexible I/O Tester development
 help / color / mirror / Atom feed
* eta broken at numjobs=1024
@ 2010-05-16  0:48 Erik Inge Bolsø
  2010-05-17 17:43 ` Jens Axboe
  0 siblings, 1 reply; 2+ messages in thread
From: Erik Inge Bolsø @ 2010-05-16  0:48 UTC (permalink / raw)
  To: fio

Or rather, eta is broken while fio is busy forking processes, which takes 
some time at numjobs=1024. And while they are shutting down at the end a 
run. (fio 1.38)

Early in the run:

[139M/15217K /s] [35K/3715 iops] [eta 1158035696d:06h:32m:18s]


At the end:

Jobs: 8 (f=8): 
[___________________________________________________________________________________________________________________________________________________________________________________________________________________m___m__________________________________mmmmmm] 
[0.5% done] [91226K/10174K /s] [22K/2484 iops] [eta 02h:04m:00s]

Hopefully just a cosmetic issue...



Using this job file:

[global]
direct=1
rw=randrw
rwmixread=90
rwmixwrite=10
bs=4k
size=10G
numjobs=1024
runtime=30
time_based
group_reporting

[rw-iops]
filename=/dev/fioa2


Btw, what's the intended replacement for rwmixcycle, since that is marked 
deprecated in the source and on use?

--
-erik	http://useofwords.blogspot.com/

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

* Re: eta broken at numjobs=1024
  2010-05-16  0:48 eta broken at numjobs=1024 Erik Inge Bolsø
@ 2010-05-17 17:43 ` Jens Axboe
  0 siblings, 0 replies; 2+ messages in thread
From: Jens Axboe @ 2010-05-17 17:43 UTC (permalink / raw)
  To: Erik Inge Bolsø; +Cc: fio

On Sun, May 16 2010, Erik Inge Bols� wrote:
> Or rather, eta is broken while fio is busy forking processes, which takes 
> some time at numjobs=1024. And while they are shutting down at the end a 
> run. (fio 1.38)
> 
> Early in the run:
> 
> [139M/15217K /s] [35K/3715 iops] [eta 1158035696d:06h:32m:18s]
> 
> 
> At the end:
> 
> Jobs: 8 (f=8): 
> [___________________________________________________________________________________________________________________________________________________________________________________________________________________m___m__________________________________mmmmmm] 
> [0.5% done] [91226K/10174K /s] [22K/2484 iops] [eta 02h:04m:00s]
> 
> Hopefully just a cosmetic issue...

It's a cosmetic issue, I'll add it to the todo list.

> Btw, what's the intended replacement for rwmixcycle, since that is marked 
> deprecated in the source and on use?

There's no replacement, the rwmixread/rwmixwrite are the only exposed
options.

-- 
Jens Axboe


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

end of thread, other threads:[~2010-05-17 17:43 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2010-05-16  0:48 eta broken at numjobs=1024 Erik Inge Bolsø
2010-05-17 17:43 ` Jens Axboe

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox