All of lore.kernel.org
 help / color / mirror / Atom feed
* Fwd: RE: FIo version 2.1.8 with write_iops_log option doesn't work as expected
       [not found] <001b01d2b403$1aa072b0$4fe15810$@citigo.com.vn>
@ 2017-04-13 15:59 ` Rebecca Cran
  2017-04-13 20:55   ` Sitsofe Wheeler
  0 siblings, 1 reply; 3+ messages in thread
From: Rebecca Cran @ 2017-04-13 15:59 UTC (permalink / raw)
  To: fio; +Cc: Son Chu

I'm seeing the same thing on Linux and Windows. Can somebody help Son?

-------- Forwarded Message --------
Subject: 	RE: FIo version 2.1.8 with write_iops_log option doesn't work 
as expected
Date: 	Thu, 13 Apr 2017 10:07:36 +0700
From: 	Son Chu <son.ct@citigo.com.vn>
To: 	'Rebecca Cran' <rebecca@bluestop.org>



Hi Rebecca,

So to get more detail, I will list what I did on both Linux and Windows 
to let you see what is my issue:

-Linux Ubuntu 16.04:

oVersion

�2.2.10

oCommand

�fio --randrepeat=1 --ioengine=libaio --direct=1 --name=test 
--filename=test --bs=4k --iodepth=64 --size=100M --readwrite=randrw 
--rwmixread=75 --write_iops_log=randrw

oOutput

�

-Windows Server 2012 R2:

oVersion

�2.1.8

oCommand

�fio.exe --randrepeat=1 --ioengine=windowsaio --direct=1 --name=test 
--filename=test --bs=4k --iodepth=64 --size=100M --readwrite=randrw 
--rwmixread=75 --write_iops_log=randrw

oOutput

33, 1, 0, 4096

34, 1, 1, 4096

35, 1, 1, 4096

44, 1, 0, 4096

46, 1, 0, 4096

53, 1, 0, 4096

54, 1, 1, 4096

58, 1, 0, 4096

63, 1, 0, 4096

64, 1, 1, 4096

64, 1, 1, 4096

73, 1, 0, 4096

81, 1, 0, 4096

89, 1, 0, 4096

98, 1, 0, 4096

 From these result I wonder why the second column from Windows Server is 
different from Linux???? It should be the IOPS like 844,289 from Linux 
than 1 from Windows.

Can you help me on this please?

Thanks a lot in advance.

Regards,

Son

*From:*Rebecca Cran [mailto:rebecca@bluestop.org]
*Sent:* Wednesday, April 12, 2017 10:53 PM
*To:* Son Chu <son.ct@citigo.com.vn>
*Subject:* Re: FIo version 2.1.8 with write_iops_log option doesn't work 
as expected

I'm seeing exactly the same on Linux. What commandline are you using there?

Rebecca

On 4/11/2017 10:28 PM, Son Chu wrote:

    Hi Rebecca,

    I am using FIO version 2.1.8 on Windows Server 2012 R2. When I was
    trying to write iops log with option write_iops_log follow this command:

    fio --directory=C\:\fio --ioengine=windowsaio --direct=1
    --name=randwin --filename=randwin --bs=4k --iodepth=16 --size=4G
    --readwrite=randrw --rwmixread=75 --write_bw_log=randwin
    --write_iops_log=randwin --write_lat_log=randwin --per_job_logs=0

    I got a log file with example content below:

    138, 1, 1, 4096

    138, 1, 1, 4096

    138, 1, 0, 4096

    139, 1, 0, 4096

    141, 1, 0, 4096

    143, 1, 0, 4096

    143, 1, 1, 4096

    145, 1, 0, 4096

    147, 1, 0, 4096

    147, 1, 1, 4096

    147, 1, 1, 4096

    149, 1, 0, 4096

    151, 1, 0, 4096

    152, 1, 0, 4096

    155, 1, 0, 4096

    158, 1, 0, 4096

    160, 1, 0, 4096

    162, 1, 0, 4096

    164, 1, 0, 4096

    165, 1, 0, 4096

    166, 1, 1, 4096

    166, 1, 1, 4096

    166, 1, 1, 4096

    167, 1, 1, 4096

    168, 1, 0, 4096

    I have wondered why the second colume was 1 (iops) instead of the expected iops per second (milisecond). If I used command with option: log_avg_msec=1000 then I got expected iops value. This is not like Linux.

      

    Can you please help on this?

      

    Sorry if this disturbs you.

      

    Thanks a lot in advance.

      

    Regards,

    Son


^ permalink raw reply	[flat|nested] 3+ messages in thread

* Re: RE: FIo version 2.1.8 with write_iops_log option doesn't work as expected
  2017-04-13 15:59 ` Fwd: RE: FIo version 2.1.8 with write_iops_log option doesn't work as expected Rebecca Cran
@ 2017-04-13 20:55   ` Sitsofe Wheeler
  2017-04-14  2:14     ` Son Chu
  0 siblings, 1 reply; 3+ messages in thread
From: Sitsofe Wheeler @ 2017-04-13 20:55 UTC (permalink / raw)
  To: Son Chu; +Cc: fio, Rebecca Cran

Hi Son,

I tried fio on Windows (fio 2.18), Linux (fio
2.19) and macOS (fio 2.19) and all produce the same results for me with
the following:

$ fio --size=10M --ioengine=null --time_based=1 --runtime=5s
--write_iops_log=iops --rate_iops=50 --name iops
[...]
$ head -5 iops_iops.1.log
0, 1, 0, 4096
0, 1, 0, 4096
42, 1, 0, 4096
60, 1, 0, 4096
80, 1, 0, 4096

In all cases the second column contained one. You didn't actually
include the output of your Linux run but you seemed to be suggesting
that the second column wasn't "1" there by default but as said I do
see 1 in that column on Linux.

If you upgrade your Linux fio to something recent like 2.19 (2.2.10 is
very old!) do you see 1 in that column by default? Perhaps the default
behaviour changed over the years...

On 13 April 2017 at 16:59, Rebecca Cran <rebecca@bluestop.org> wrote:
> I'm seeing the same thing on Linux and Windows. Can somebody help Son?
>
> -------- Forwarded Message --------
> Subject:        RE: FIo version 2.1.8 with write_iops_log option doesn't
> work as expected
> Date:   Thu, 13 Apr 2017 10:07:36 +0700
> From:   Son Chu <son.ct@citigo.com.vn>
> To:     'Rebecca Cran' <rebecca@bluestop.org>
>
>
>
> Hi Rebecca,
>
> So to get more detail, I will list what I did on both Linux and Windows to
> let you see what is my issue:
>
> -Linux Ubuntu 16.04:
>
> oVersion
>
> §2.2.10
>
> oCommand
>
> §fio --randrepeat=1 --ioengine=libaio --direct=1 --name=test --filename=test
> --bs=4k --iodepth=64 --size=100M --readwrite=randrw --rwmixread=75
> --write_iops_log=randrw
>
> oOutput
>
> §
>
> -Windows Server 2012 R2:
>
> oVersion
>
> §2.1.8
>
> oCommand
>
> §fio.exe --randrepeat=1 --ioengine=windowsaio --direct=1 --name=test
> --filename=test --bs=4k --iodepth=64 --size=100M --readwrite=randrw
> --rwmixread=75 --write_iops_log=randrw
>
> oOutput
>
> 33, 1, 0, 4096
>
> 34, 1, 1, 4096
>
> 35, 1, 1, 4096
>
> 44, 1, 0, 4096
>
> 46, 1, 0, 4096
>
> 53, 1, 0, 4096
>
> 54, 1, 1, 4096
>
> 58, 1, 0, 4096
>
> 63, 1, 0, 4096
>
> 64, 1, 1, 4096
>
> 64, 1, 1, 4096
>
> 73, 1, 0, 4096
>
> 81, 1, 0, 4096
>
> 89, 1, 0, 4096
>
> 98, 1, 0, 4096
>
> From these result I wonder why the second column from Windows Server is
> different from Linux???? It should be the IOPS like 844,289 from Linux than
> 1 from Windows.
>
> Can you help me on this please?
>
> Thanks a lot in advance.
>
> Regards,
>
> Son
>
> *From:*Rebecca Cran [mailto:rebecca@bluestop.org]
> *Sent:* Wednesday, April 12, 2017 10:53 PM
> *To:* Son Chu <son.ct@citigo.com.vn>
> *Subject:* Re: FIo version 2.1.8 with write_iops_log option doesn't work as
> expected
>
> I'm seeing exactly the same on Linux. What commandline are you using there?
>
> Rebecca
>
> On 4/11/2017 10:28 PM, Son Chu wrote:
>
>    Hi Rebecca,
>
>    I am using FIO version 2.1.8 on Windows Server 2012 R2. When I was
>    trying to write iops log with option write_iops_log follow this command:
>
>    fio --directory=C\:\fio --ioengine=windowsaio --direct=1
>    --name=randwin --filename=randwin --bs=4k --iodepth=16 --size=4G
>    --readwrite=randrw --rwmixread=75 --write_bw_log=randwin
>    --write_iops_log=randwin --write_lat_log=randwin --per_job_logs=0
>
>    I got a log file with example content below:
>
>    138, 1, 1, 4096
>
>    138, 1, 1, 4096
>
>    138, 1, 0, 4096
>
>    139, 1, 0, 4096
>
>    141, 1, 0, 4096
>
>    143, 1, 0, 4096
>
>    143, 1, 1, 4096
>
>    145, 1, 0, 4096
>
>    147, 1, 0, 4096
>
>    147, 1, 1, 4096
>
>    147, 1, 1, 4096
>
>    149, 1, 0, 4096
>
>    151, 1, 0, 4096
>
>    152, 1, 0, 4096
>
>    155, 1, 0, 4096
>
>    158, 1, 0, 4096
>
>    160, 1, 0, 4096
>
>    162, 1, 0, 4096
>
>    164, 1, 0, 4096
>
>    165, 1, 0, 4096
>
>    166, 1, 1, 4096
>
>    166, 1, 1, 4096
>
>    166, 1, 1, 4096
>
>    167, 1, 1, 4096
>
>    168, 1, 0, 4096
>
>    I have wondered why the second colume was 1 (iops) instead of the
> expected iops per second (milisecond). If I used command with option:
> log_avg_msec=1000 then I got expected iops value. This is not like Linux.
>
>
>    Can you please help on this?
>
>
>    Sorry if this disturbs you.
>
>
>    Thanks a lot in advance.
>
>
>    Regards,
>
>    Son

-- 
Sitsofe | http://sucs.org/~sits/

^ permalink raw reply	[flat|nested] 3+ messages in thread

* RE: RE: FIo version 2.1.8 with write_iops_log option doesn't work as expected
  2017-04-13 20:55   ` Sitsofe Wheeler
@ 2017-04-14  2:14     ` Son Chu
  0 siblings, 0 replies; 3+ messages in thread
From: Son Chu @ 2017-04-14  2:14 UTC (permalink / raw)
  To: 'Sitsofe Wheeler'; +Cc: fio, 'Rebecca Cran'

Hi Sitsofe,

Thank you very much for your help. I got the same result as you and Rebecca said after upgrade fio to version 2.19 . My version was too old (2.2.10). 

Thank you all for the helping.

Best Regards,
Son

-----Original Message-----
From: Sitsofe Wheeler [mailto:sitsofe@gmail.com] 
Sent: Friday, April 14, 2017 3:56 AM
To: Son Chu <son.ct@citigo.com.vn>
Cc: fio@vger.kernel.org; Rebecca Cran <rebecca@bluestop.org>
Subject: Re: RE: FIo version 2.1.8 with write_iops_log option doesn't work as expected

Hi Son,

I tried fio on Windows (fio 2.18), Linux (fio
2.19) and macOS (fio 2.19) and all produce the same results for me with
the following:

$ fio --size=10M --ioengine=null --time_based=1 --runtime=5s
--write_iops_log=iops --rate_iops=50 --name iops
[...]
$ head -5 iops_iops.1.log
0, 1, 0, 4096
0, 1, 0, 4096
42, 1, 0, 4096
60, 1, 0, 4096
80, 1, 0, 4096

In all cases the second column contained one. You didn't actually
include the output of your Linux run but you seemed to be suggesting
that the second column wasn't "1" there by default but as said I do
see 1 in that column on Linux.

If you upgrade your Linux fio to something recent like 2.19 (2.2.10 is
very old!) do you see 1 in that column by default? Perhaps the default
behaviour changed over the years...

On 13 April 2017 at 16:59, Rebecca Cran <rebecca@bluestop.org> wrote:
> I'm seeing the same thing on Linux and Windows. Can somebody help Son?
>
> -------- Forwarded Message --------
> Subject:        RE: FIo version 2.1.8 with write_iops_log option doesn't
> work as expected
> Date:   Thu, 13 Apr 2017 10:07:36 +0700
> From:   Son Chu <son.ct@citigo.com.vn>
> To:     'Rebecca Cran' <rebecca@bluestop.org>
>
>
>
> Hi Rebecca,
>
> So to get more detail, I will list what I did on both Linux and Windows to
> let you see what is my issue:
>
> -Linux Ubuntu 16.04:
>
> oVersion
>
> §2.2.10
>
> oCommand
>
> §fio --randrepeat=1 --ioengine=libaio --direct=1 --name=test --filename=test
> --bs=4k --iodepth=64 --size=100M --readwrite=randrw --rwmixread=75
> --write_iops_log=randrw
>
> oOutput
>
> §
>
> -Windows Server 2012 R2:
>
> oVersion
>
> §2.1.8
>
> oCommand
>
> §fio.exe --randrepeat=1 --ioengine=windowsaio --direct=1 --name=test
> --filename=test --bs=4k --iodepth=64 --size=100M --readwrite=randrw
> --rwmixread=75 --write_iops_log=randrw
>
> oOutput
>
> 33, 1, 0, 4096
>
> 34, 1, 1, 4096
>
> 35, 1, 1, 4096
>
> 44, 1, 0, 4096
>
> 46, 1, 0, 4096
>
> 53, 1, 0, 4096
>
> 54, 1, 1, 4096
>
> 58, 1, 0, 4096
>
> 63, 1, 0, 4096
>
> 64, 1, 1, 4096
>
> 64, 1, 1, 4096
>
> 73, 1, 0, 4096
>
> 81, 1, 0, 4096
>
> 89, 1, 0, 4096
>
> 98, 1, 0, 4096
>
> From these result I wonder why the second column from Windows Server is
> different from Linux???? It should be the IOPS like 844,289 from Linux than
> 1 from Windows.
>
> Can you help me on this please?
>
> Thanks a lot in advance.
>
> Regards,
>
> Son
>
> *From:*Rebecca Cran [mailto:rebecca@bluestop.org]
> *Sent:* Wednesday, April 12, 2017 10:53 PM
> *To:* Son Chu <son.ct@citigo.com.vn>
> *Subject:* Re: FIo version 2.1.8 with write_iops_log option doesn't work as
> expected
>
> I'm seeing exactly the same on Linux. What commandline are you using there?
>
> Rebecca
>
> On 4/11/2017 10:28 PM, Son Chu wrote:
>
>    Hi Rebecca,
>
>    I am using FIO version 2.1.8 on Windows Server 2012 R2. When I was
>    trying to write iops log with option write_iops_log follow this command:
>
>    fio --directory=C\:\fio --ioengine=windowsaio --direct=1
>    --name=randwin --filename=randwin --bs=4k --iodepth=16 --size=4G
>    --readwrite=randrw --rwmixread=75 --write_bw_log=randwin
>    --write_iops_log=randwin --write_lat_log=randwin --per_job_logs=0
>
>    I got a log file with example content below:
>
>    138, 1, 1, 4096
>
>    138, 1, 1, 4096
>
>    138, 1, 0, 4096
>
>    139, 1, 0, 4096
>
>    141, 1, 0, 4096
>
>    143, 1, 0, 4096
>
>    143, 1, 1, 4096
>
>    145, 1, 0, 4096
>
>    147, 1, 0, 4096
>
>    147, 1, 1, 4096
>
>    147, 1, 1, 4096
>
>    149, 1, 0, 4096
>
>    151, 1, 0, 4096
>
>    152, 1, 0, 4096
>
>    155, 1, 0, 4096
>
>    158, 1, 0, 4096
>
>    160, 1, 0, 4096
>
>    162, 1, 0, 4096
>
>    164, 1, 0, 4096
>
>    165, 1, 0, 4096
>
>    166, 1, 1, 4096
>
>    166, 1, 1, 4096
>
>    166, 1, 1, 4096
>
>    167, 1, 1, 4096
>
>    168, 1, 0, 4096
>
>    I have wondered why the second colume was 1 (iops) instead of the
> expected iops per second (milisecond). If I used command with option:
> log_avg_msec=1000 then I got expected iops value. This is not like Linux.
>
>
>    Can you please help on this?
>
>
>    Sorry if this disturbs you.
>
>
>    Thanks a lot in advance.
>
>
>    Regards,
>
>    Son

-- 
Sitsofe | http://sucs.org/~sits/


^ permalink raw reply	[flat|nested] 3+ messages in thread

end of thread, other threads:[~2017-04-14  3:17 UTC | newest]

Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
     [not found] <001b01d2b403$1aa072b0$4fe15810$@citigo.com.vn>
2017-04-13 15:59 ` Fwd: RE: FIo version 2.1.8 with write_iops_log option doesn't work as expected Rebecca Cran
2017-04-13 20:55   ` Sitsofe Wheeler
2017-04-14  2:14     ` Son Chu

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.