On 2016-12-28 15:41, Paul Neuwirth wrote:
After resuming to solve my problem, I noticed during comparing of outputs of btrfs checks, that each run reported quite random errors. mdadm never ever reported any problem, there was no unclean stop/unmount or shutdown at all. But running a check of the RAID 1 array, there are mismatches (/sys/block/md2/md/mismatch_cnt shows around 15k at 20% progress). smartctl not showing any problems, repeated selftests are ok.
I suppose that mdadm doesn't report a problem because it is working on it. It is not an issue yet. maybe btrfs is more sensitive.
What can have caused this? I am still wondering if this "flush" can cause this:
Maybe. I don't know. Some people say that consumer class hard disks are not really raid ready. Consider this post, for instance: https://lists.opensuse.org/opensuse/2016-12/msg01097.html There have been several threads related to raid this month. Maybe you should have a look for ideas. -- Cheers / Saludos, Carlos E. R. (from 42.2 x86_64 "Malachite" at Telcontar)