On 03/17/2016 11:01 PM, 王 铮 wrote:
Hi all, I am now using FIO2.8, but it seems the parm "loops = int" do not works. The following is my script. I expected to do a test with 25 loops 32K SEQ Read and each loop run 10 seconds. But the result is that the test only ran once and then reported the data. Can any one help on this? Thank you ! Script code: [global] filename=/dev/nvme0n1 # test device ioengine=libaio # Default engine for CentOS disable_slat=1 # disables Submission Latency recording disable_lat=1 # disables Total Latency recording direct=1 # Uses direct IO's, bypass OS buffer ba=4k # all IO's 4K aligned size=100% # numjobs=12 # Num of jobs(accdroing to CPU threads) norandommap group_reporting [32K_SEQ_Write_Q1] # Job name stonewall bs=32K # block transfer size rw=read # r/w iodepth=256 loops=25 runtime=10
It's because you set runtime=10, that takes precedence over the number of loops. So the job will do 25 loops, if it can do 25 loops within the 10 seconds. If you want 25 loops, then it would probably be easier to limit the amount of IO done per loop instead. If you did:
loops=25 io_limit=10g then you would get 25 loops, each doing 10g of IO. -- Jens Axboe -- 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
