On Tue, Nov 5, 2019 at 2:13 PM Sitsofe Wheeler <sitsofe@xxxxxxxxx> wrote: > > My understanding is that the md "chunk size" is 64k so basically > you're saying the sectsz should have been manually set to be as big as > possible at mkfs time? I never realised this never happened by default > (I see the sunit seems to be correct given the block size of 4096 but > I'm unsure about swidth)... Check with # mdadm -E /dev/sdXY Chunk size depends on the version of mdadm at the time of creation. It's been a while since 512KiB was the default, leading to rather large full stripe write size, and a lot of RMW if a significant number of writes aren't full stripe writes. -- Chris Murphy