From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail-ie0-f175.google.com ([209.85.223.175]:64181 "EHLO mail-ie0-f175.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754829AbaHEQFk (ORCPT ); Tue, 5 Aug 2014 12:05:40 -0400 Received: by mail-ie0-f175.google.com with SMTP id x19so1262825ier.6 for ; Tue, 05 Aug 2014 09:05:39 -0700 (PDT) MIME-Version: 1.0 In-Reply-To: <9E8FC62FFD81D74BA20610CD2B890B843DF4F8@SHSMSX101.ccr.corp.intel.com> References: <9E8FC62FFD81D74BA20610CD2B890B843DF4F8@SHSMSX101.ccr.corp.intel.com> Date: Tue, 5 Aug 2014 12:05:39 -0400 Message-ID: Subject: Re: bug report From: Vasily Tarasov Content-Type: text/plain; charset=UTF-8 Sender: fio-owner@vger.kernel.org List-Id: fio@vger.kernel.org To: "Zhu, Zhengyun" Cc: "fio@vger.kernel.org" Hi Zhu, If the total run time of the job is only 30 seconds, there is no way to have a warm-up time more than 30 seconds. In fact, during the warm-up time fio executes exactly same workload as during the "main" run. That is a definition of warm-up time. So, if the whole workload can run for only 30 seconds, there is no way to warm the system up for more than 30 seconds with this workload. I think you need to make your experiment longer. For example, you can use "loops=" parameter to execute a job multiple times. HTH, Vasily On Tue, Aug 5, 2014 at 1:08 AM, Zhu, Zhengyun wrote: > > Hi, > > When I am using fio to do some tests, I find a problem. > > I am using a client with a 10Gb NIC to do the test case of sequential read, and the configuration file is below: > > //version 2.0.8 > global] > iodepth=256 > runtime=600 > ioengine=libaio > direct=1 > size=30g > filename=/dev/vdb > ramp_time=10 > > [seq-read-64k] > rw=read > bs=64k > iodepth_batch_submit=8 > iodepth_batch_complete=8 > > > we know the bandwidth can reach nearly 1100MB/s without the parameter of rate. It will finish fio in about 30 seconds, because the size is 30g. > > first, when I set ramp_time to 10, the fio result is below: > > seq-read-64k: (groupid=0, jobs=1): err= 0: pid=26840 > read : io=20016MB, bw=1105.8MB/s, iops=17677 , runt= 18102msec > > second, when I set ramp_time to 35, the fio result is below: > > seq-read-64k: (groupid=0, jobs=1): err= 0: pid=26861 > read : io=30720MB, bw=1102.3MB/s, iops=17636 , runt= 27870msec > > According to these, we can summary: > > 1. When we set ramp_time less than the time of fio finish read(we call it fftime for short), then > Runt = fftime - ramptime > 2. When we set ramp_time greater than the time of fio finish read(we call it fftime for short), then > Runt = fftime > My requirement is: > I want to use ramp time to warm up 50 seconds, and then collect iostat data, but fio has finished after 30 seconds, so I will collect all empty data. > > Could we not calculate time and io size during ramp time in runt time, or do I have some mistakes to understand it? > -- > To unsubscribe from this list: send the line "unsubscribe fio" in > the body of a message to majordomo@vger.kernel.org > More majordomo info at http://vger.kernel.org/majordomo-info.html >