On Sun, Apr 16, 2023 at 10:20:45PM -0700, Kyle Sanderson wrote: > The single btrfs disk was at 100% utilization and a wa of 50~, reading > back at around 2MB/s. df and similar would simply freeze. Leading up > to this I removed around 2T of data from a single btrfs disk. I > managed to get most of the services shutdown and disks unmounted, but > when the system came back up I had to use xfs_repair (for the first What exactly was the error messages XFS emitted when it failed to mount, and what did xfs_repair fix to enable it to boot? Unless you have a broken disk (i.e. firmware either lies about cache flush completion or the implementation is broken), there is no reason for any amount of load on the disk to cause filesystem corruption.... -Dave. -- Dave Chinner david@xxxxxxxxxxxxx
