Hi, On Sat, 11 Jul 2020 at 01:33, siha lawrence <sihaj33@xxxxxxxxx> wrote: > > Below is the workload we have run and we see PHY changes during I/O > Does fio abort outstanding commands? No fio doesn't (I believe there's an outstanding request to add functionality to do so - https://github.com/axboe/fio/issues/292 ) but something outside of fio might be... I guess you're scribbling over the start and end of the disk and it's not impossible you're triggering udev rules that try and re-read partitions when areas that might contain partition tables are changed. You can try fiddling with --offset so you start 10MBytes in and change size to avoid writing the very end of the disk to try and narrow something like that... Let us know if you find out the cause! > > Workload: > fio --eta=never --ioengine=libaio --size=100% --ramp_time=10 > --runtime=40 --time_based=1 --verify_backlog=1 --verify_dump=1 > --verify_fatal=1 --norandommap --numjobs=1 --direct=1 > --random_generator=tausworthe64 --iodepth=8 --rw=write --bs=1M > --buffer_pattern=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 > --minimal --name=/dev/sdp --name=/dev/sdq --name=/dev/sdr > --name=/dev/sds --name=/dev/sdt --name=/dev/sdu --name=/dev/sdv > --name=/dev/sdw --name=/dev/sdx --name=/dev/sdy --name=/dev/sdz > --name=/dev/sdaa --name=/dev/sdaf --name=/dev/sdag --name=/dev/sdah > --name=/dev/sdai --name=/dev/sdab --name=/dev/sdac --name=/dev/sdad > --name=/dev/sdae --name=/dev/sdaj --name=/dev/sdak --name=/dev/sdal > --name=/dev/sdam --name=/dev/sdan --name=/dev/sdao --name=/dev/sdaq > --name=/dev/sdap --name=/dev/sdf --name=/dev/sdg --name=/dev/sdh > --name=/dev/sdi --name=/dev/sdj --name=/dev/sdk --name=/dev/sdl > --name=/dev/sdm --name=/dev/sdo --name=/dev/sdn --name=/dev/sde > --name=/dev/sdd --name=/dev/sdc --name=/dev/sdb > > We do not see this behaviour with SAS drives. Maybe different devices trigger different rules? -- Sitsofe | http://sucs.org/~sits/
