json+ max latency logging for bins

json+ max latency logging for bins

[abhishek koundal]

Can someone confirm if json+ format is used the latency reported in
the bins (~1200) are mid-point not the max?

 Does the --status-interval=1sec makes reporting average value in bins.

Does adding the log_max_value=1 , will ensure that always max is
recorded in the latency bin? I have used this for the non-json format
but then average latency reporting per interval is lost.

Appreciate your response .

-- 
Life is too short for silly things so invest your time in some
productive outputs!!

Re: json+ max latency logging for bins

[Sitsofe Wheeler]

Hi,

On 28 September 2017 at 02:04, abhishek koundal <akoundal@gmail.com> wrote:
> Can someone confirm if json+ format is used the latency reported in
> the bins (~1200) are mid-point not the max?

There's a brief description of the JSON+ format in the latest git
version of the HOWTO (see the JSON+ output section of the HOWTO
http://fio.readthedocs.io/en/latest/fio_doc.html#id3 ).

>  Does the --status-interval=1sec makes reporting average value in bins.

Averaged over what period? Bear in mind even with
--status-interval=<time> fio is reporting cumulative data since the
job started running ...

> Does adding the log_max_value=1 , will ensure that always max is
> recorded in the latency bin? I have used this for the non-json format
> but then average latency reporting per interval is lost.

I don't know for sure but I'd guess that value ONLY affects the logs
produced by *_bw_log and not the JSON outputs.

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