Thank you, I searched high and low and missed that thread.
David Byte
Sr. Technology Strategist
Alliances and SUSE Embedded
dbyte@xxxxxxxx
918.528.4422
On 6/13/17, 3:57 PM, "Sitsofe Wheeler" <sitsofe@xxxxxxxxx> wrote:
On 13 June 2017 at 15:04, David Byte <dbyte@xxxxxxxx> wrote:
>
> I tried this with both 2.13 (the version that is available with SUSE Enterprise Storage and with a freshly built 2.20-53 that I built 12 hours ago. Neither prints the actual iops value to the iops log. Any clue as to where the problem lies?
>
> Raw data follows.
>
> linux:~/bench # ./fio --filename=/root/test.fil --size=1G --runtime=30 --rw=randwrite --write_iops_log=4kw3 --name=test
> test: (g=0): rw=randwrite, bs=(R) 4096B-4096B, (W) 4096B-4096B, (T) 4096B-4096B, ioengine=psync, iodepth=1
> fio-2.20-53-g85c7
> Starting 1 process
> Jobs: 1 (f=1): [w(1)][100.0%][r=0KiB/s,w=30.5MiB/s][r=0,w=7802 IOPS][eta 00m:00s]
> test: (groupid=0, jobs=1): err= 0: pid=4362: Fri Jun 9 19:54:13 2017
> write: IOPS=8563, BW=33.5MiB/s (35.1MB/s)(1004MiB/30001msec)
[...]
> linux:~/bench # tail 4kw3_iops.1.log
> 29986, 1, 1, 4096
> 29986, 1, 1, 4096
> 29986, 1, 1, 4096
> 29986, 1, 1, 4096
That looks like it's printing logging information for each individual I/O
rather than averaging them over a given period (so you've gained
precision but you'd have to post process the result to work out the
IOPS). In fact this looks like a similar question to that of another
SUSE engineer (who was looking at Ceph) over on
https://github.com/axboe/fio/issues/337 ...
Just to check, have you seen the log_avg_msec option
(http://fio.readthedocs.io/en/latest/fio_doc.html#cmdoption-arg-log-avg-msec
) and are you aware it defaults to 0?
--
Sitsofe | http://sucs.org/~sits/
��.n��������+%������w��{.n�������^n�r������&��z�ޗ�zf���h���~����������_��+v���)ߣ�
