Has anyone run into this kind of behavior before? I did a distro update from 15.4 to 15.5 and decided to take a shot at installing the kernel backports. Things seemed to run fine for a few days, but over the last couple of days I’ve run into IO errors reported by docker and reported corruption of some files on my main 5-disk raid-5 array. No issues on the system disk reported or on my smaller 2-disk raid-1 array.
Then this morning I was greeted by this error for my daily balance job
ERROR: error during balancing '/mnt2/Media': Input/output error
This is the balance command in cron, flock is just there to make sure no other jobs kick off if it is doing a balance/scrub/etc.
flock -w 0 -x /tmp/btrfs.lck btrfs balance start -dusage=85 -dlimit=10 -musage=85 -mlimit=10 /mnt2/Media
Manually performing a scrub after this error message dropped a lot of corrupt file messages into dmesg, so I reverted back to kernel 5.14.21-150500.55.59 and everything appears to be running fine (all docker containers are online and reporting healthy) and a subsequent scrub has not found any errors at the moment. It’ll take another ~60 hours to complete, but that’s on par with what my monthly scrubs take.
Has anyone else seen this behavior? Thoughts on what could cause it?
BTRFS progs is the latest version from the OpenSuse filesystems repo.