On 12/13/18 3:19 PM, Matthew Eaton wrote: > Using ramp_time is causing all min values (slat/clat/lat/bw/iops) to report 0. > Removing ramp_time fixes the issue. > > Job file: > [rnd-8020] > ioengine=libaio > direct=1 > bs=4k > rw=randrw > rwmixread=80 > numjobs=1 > iodepth=32 > runtime=10 > time_based > norandommap > randrepeat=0 > filename=/dev/sdb > #ramp_time=10 > > Fio output with min values zeroed: > rnd-8020: (g=0): rw=randrw, bs=(R) 4096B-4096B, (W) 4096B-4096B, (T) > 4096B-4096B, ioengine=libaio, iodepth=32 > fio-3.12-45-g771c > Starting 1 process > Jobs: 1 (f=1): [m(1)][100.0%][r=149MiB/s,w=37.4MiB/s][r=38.1k,w=9562 > IOPS][eta 00m:00s] > rnd-8020: (groupid=0, jobs=1): err= 0: pid=4701: Thu Dec 13 14:14:36 2018 > read: IOPS=33.6k, BW=131MiB/s (138MB/s)(1313MiB/10002msec) > slat (nsec): min=0, max=14301, avg=1880.55, stdev=242.49 > clat (nsec): min=0, max=9948.1k, avg=633109.98, stdev=597207.47 > lat (nsec): min=0, max=9950.0k, avg=635050.69, stdev=597212.18 > clat percentiles (usec): > | 1.00th=[ 135], 5.00th=[ 165], 10.00th=[ 178], 20.00th=[ 202], > | 30.00th=[ 227], 40.00th=[ 251], 50.00th=[ 285], 60.00th=[ 379], > | 70.00th=[ 873], 80.00th=[ 1270], 90.00th=[ 1565], 95.00th=[ 1762], > | 99.00th=[ 2212], 99.50th=[ 2606], 99.90th=[ 3851], 99.95th=[ 4015], > | 99.99th=[ 5211] > bw ( KiB/s): min= 0, max=164608, per=100.00%, avg=134473.20, > stdev=12836.10, samples=20 > iops : min= 0, max=41152, avg=33618.30, stdev=3209.03, samples=20 > write: IOPS=8419, BW=32.9MiB/s (34.5MB/s)(329MiB/10002msec); 0 zone resets > slat (nsec): min=0, max=13697, avg=2059.16, stdev=383.41 > clat (nsec): min=0, max=10090k, avg=1261001.47, stdev=575428.55 > lat (nsec): min=0, max=10092k, avg=1263121.22, stdev=575424.23 > clat percentiles (usec): > | 1.00th=[ 474], 5.00th=[ 635], 10.00th=[ 709], 20.00th=[ 807], > | 30.00th=[ 898], 40.00th=[ 996], 50.00th=[ 1106], 60.00th=[ 1237], > | 70.00th=[ 1418], 80.00th=[ 1680], 90.00th=[ 2089], 95.00th=[ 2376], > | 99.00th=[ 3064], 99.50th=[ 3523], 99.90th=[ 4424], 99.95th=[ 4752], > | 99.99th=[ 5866] > bw ( KiB/s): min= 0, max=40240, per=100.00%, avg=33682.40, > stdev=3069.64, samples=20 > iops : min= 0, max=10060, avg=8420.60, stdev=767.41, samples=20 > lat (usec) : 100=0.20%, 250=31.37%, 500=20.39%, 750=5.34%, 1000=8.79% > lat (msec) : 2=30.07%, 4=3.76%, 10=0.08%, 20=0.01% > cpu : usr=9.58%, sys=12.38%, ctx=406871, majf=0, minf=15 > IO depths : 1=0.0%, 2=0.0%, 4=0.0%, 8=0.0%, 16=0.0%, 32=100.0%, >=64=0.0% > submit : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0% > complete : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.1%, 64=0.0%, >=64=0.0% > issued rwts: total=336181,84209,0,0 short=0,0,0,0 dropped=0,0,0,0 > latency : target=0, window=0, percentile=100.00%, depth=32 Huh, that's no good. Pushed a fix: http://git.kernel.dk/cgit/fio/commit/?id=5b8f19b7afe0cabc002c453a1a4abd7a494880bb -- Jens Axboe