Thanks, it was the missing time_based indeed. Regards, Andrey On Mon, Mar 5, 2018 at 12:25 PM, Sitsofe Wheeler <sitsofe@xxxxxxxxx> wrote: > On 5 March 2018 at 09:14, Andrey Kuzmin <andrey.v.kuzmin@xxxxxxxxx> wrote: >> >> >> On Mon, Mar 5, 2018, 12:11 Sitsofe Wheeler <sitsofe@xxxxxxxxx> wrote: >>> >>> Hi, >>> >>> On 4 March 2018 at 21:50, Andrey Kuzmin <andrey.v.kuzmin@xxxxxxxxx> wrote: >>> > >>> > I'm not using verify, so no idea re your question. That's the config >>> > in question, with some (irrelevant) omissions. >>> >>> Thanks for including this - without it answering this type of question >>> becomes a short in the dark. >>> >>> > [global] >>> > thread=1 >>> > group_reporting=1 >>> > direct=1 >>> > verify=0 >>> > ramp_time=0 >>> > >>> > [precond] >>> > numjobs=1 >>> > iodepth=128 >>> > rw=write >>> > bs=128k >>> > size=4g >>> > >>> > [randrw] >>> > stonewall >>> > numjobs=4 >>> > iodepth=32 >>> > rw=randrw >>> > bs=4k >>> > runtime=1m >>> >> >>> >> -----Original Message----- >>> >> From: fio-owner@xxxxxxxxxxxxxxx [mailto:fio-owner@xxxxxxxxxxxxxxx] On >>> >> Behalf Of Andrey Kuzmin >>> >> >>> >> I see a run of the subj resulting in the second, time-based job (which >>> >> is stonewalled to start after the size-based one) ending prematurely, with >>> >> its lifetime apparently consumed by the size-based job. This used to work >>> >> just a couple of years back ;). Am I missing something? >>> >>> runtime >>> (http://fio.readthedocs.io/en/latest/fio_doc.html#cmdoption-arg-runtime >>> ) by itself sets the maximum amount of time a job will run for but it >>> can exit earlier for other reasons (e.g. file size has been done). >>> Perhaps you also wanted to add time_based >>> (http://fio.readthedocs.io/en/latest/fio_doc.html#cmdoption-arg-time-based >>> ) ? That option says that the I/O should be looped if necessary until >>> runtime is exceeded. >> >> >> I believe it's the same with time_based, although I'll double check later >> today. May be the size-based one should use io_size instead of size? > > io_size is for controlling the amount I/O done without also messing > with the start-end region (see > http://fio.readthedocs.io/en/latest/fio_doc.html#cmdoption-arg-io-size > ). In your case I'd be surprised if it made any difference in your > first job because you aren't changing the start offset and you're > doing sequential I/O... > > -- > Sitsofe | http://sucs.org/~sits/ -- To unsubscribe from this list: send the line "unsubscribe fio" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
