Description of problem: When user creates thin-pool with unused thin LV, or caches a volume, then every call a flushing 'dmsetup status' cause very small i/o batch to pass on actual real device. This is not a big deal in case performance, but it prevents unused 'devices' to stay in sleep eventually wakes them up whenever admin runs: dmsetup status. As a workaround to keep disks sleeping - user may 'add --noflush' arg... Version-Release number of selected component (if applicable): 4.12.0-0.rc6.git3.2.fc27.x86_64 How reproducible: Steps to Reproduce: 1. create thinLV (lvcreate -L10 -V10 -T vg/pool -n thinLV) (or cache some LV) 2. iostat -d 1 3. observe i/o load with every 'dmsetup status' call Actual results: Expected results: Additional info: