Unexpected ramp_time behavior?

Unexpected ramp_time behavior?

[Devon Yablonski]

Hi, I'm using fio 2.1.7 and I have the following fio configuration:

#FIO CONFIGURATION
[globals]
name=fio_write
rw=write
size=5G
ioengine=libaio
direct=1
unlink=0
bs=4096
iodepth=128
scramble_buffers=1
group_reporting
gtod_reduce=0
norandommap
thread
randrepeat=0
ramp_time=0s
random_generator=tausworthe

[thread 0]
filename=/dev/md0

My result shows:
/fio fioc.cfg
fio_write: (g=0): rw=write, bs=4K-4K/4K-4K/4K-4K, ioengine=libaio, iodepth=128
fio-2.1.7
Starting 1 thread
Jobs: 1 (f=1): [W] [100.0% done] [0KB/799.8MB/0KB /s] [0/205K/0 iops]
[eta 00m:00s]
fio_write: (groupid=0, jobs=1): err= 0: pid=13889: Mon May 19 15:29:55 2014
  write: io=5120.0MB, bw=821510KB/s, iops=205377, runt=  6382msec
<snip>
Run status group 0 (all jobs):
  WRITE: io=5120.0MB, aggrb=821510KB/s, minb=821510KB/s,
maxb=821510KB/s, mint=6382msec, maxt=6382msec

Note that the runtime was about 6 seconds.

If I change the ramp_time to 1s (ramp_time=1s) the test runs for 12-13
seconds (about 2x) and writes much more data than size:

fio_write: (g=0): rw=write, bs=4K-4K/4K-4K/4K-4K, ioengine=libaio, iodepth=128
fio-2.1.7
Starting 1 thread
Jobs: 1 (f=1): [W] [100.0% done] [0KB/626.1MB/0KB /s] [0/160K/0 iops]
[eta 00m:00s]
fio_write: (groupid=0, jobs=1): err= 0: pid=15860: Tue May 20 08:28:35 2014
  write: io=9406.3MB, bw=717344KB/s, iops=179326, runt= 13427msec
<snip>
Run status group 0 (all jobs):
  WRITE: io=9406.3MB, aggrb=717343KB/s, minb=717343KB/s,
maxb=717343KB/s, mint=13427msec, maxt=13427msec

The time remaining counts down to 0 and then the test keeps running -
it seems like the test may be running a second iteration. What am I
doing incorrectly?

Thanks,
Devon

Re: Unexpected ramp_time behavior?

[Devon Yablonski]

Hi, does anyone have any ideas about this ramp_time behavior?

On Tue, May 20, 2014 at 10:13 AM, Devon Yablonski <dmyablonski@gmail.com> wrote:
> Hi, I'm using fio 2.1.7 and I have the following fio configuration:
>
> #FIO CONFIGURATION
> [globals]
> name=fio_write
> rw=write
> size=5G
> ioengine=libaio
> direct=1
> unlink=0
> bs=4096
> iodepth=128
> scramble_buffers=1
> group_reporting
> gtod_reduce=0
> norandommap
> thread
> randrepeat=0
> ramp_time=0s
> random_generator=tausworthe
>
> [thread 0]
> filename=/dev/md0
>
> My result shows:
> /fio fioc.cfg
> fio_write: (g=0): rw=write, bs=4K-4K/4K-4K/4K-4K, ioengine=libaio, iodepth=128
> fio-2.1.7
> Starting 1 thread
> Jobs: 1 (f=1): [W] [100.0% done] [0KB/799.8MB/0KB /s] [0/205K/0 iops]
> [eta 00m:00s]
> fio_write: (groupid=0, jobs=1): err= 0: pid=13889: Mon May 19 15:29:55 2014
>   write: io=5120.0MB, bw=821510KB/s, iops=205377, runt=  6382msec
> <snip>
> Run status group 0 (all jobs):
>   WRITE: io=5120.0MB, aggrb=821510KB/s, minb=821510KB/s,
> maxb=821510KB/s, mint=6382msec, maxt=6382msec
>
> Note that the runtime was about 6 seconds.
>
> If I change the ramp_time to 1s (ramp_time=1s) the test runs for 12-13
> seconds (about 2x) and writes much more data than size:
>
> fio_write: (g=0): rw=write, bs=4K-4K/4K-4K/4K-4K, ioengine=libaio, iodepth=128
> fio-2.1.7
> Starting 1 thread
> Jobs: 1 (f=1): [W] [100.0% done] [0KB/626.1MB/0KB /s] [0/160K/0 iops]
> [eta 00m:00s]
> fio_write: (groupid=0, jobs=1): err= 0: pid=15860: Tue May 20 08:28:35 2014
>   write: io=9406.3MB, bw=717344KB/s, iops=179326, runt= 13427msec
> <snip>
> Run status group 0 (all jobs):
>   WRITE: io=9406.3MB, aggrb=717343KB/s, minb=717343KB/s,
> maxb=717343KB/s, mint=13427msec, maxt=13427msec
>
> The time remaining counts down to 0 and then the test keeps running -
> it seems like the test may be running a second iteration. What am I
> doing incorrectly?
>
> Thanks,
> Devon

Re: Unexpected ramp_time behavior?

[Jens Axboe]

On 06/16/2014 09:07 AM, Devon Yablonski wrote:
> Hi, does anyone have any ideas about this ramp_time behavior?

I'll take a look at this, but probably not until end this week.

-- 
Jens Axboe