On Jan 22, 2014, at 11:41 AM, G. Michael Carter <mi...@carterfamily.ca> wrote:
> How do I get around this. The drive /dev/sdf has bad sectors. > > Label: Store_01 uuid: ae612523-63cf-4860-a2cb-83a26d907e43 > Total devices 5 FS bytes used 7.51TiB > devid 1 size 0.00 used 77.00GiB path /dev/sdf size 0.00 used 77.00 GB? Does this even make sense? > devid 3 size 1.82TiB used 1.41TiB path /dev/sdd > devid 4 size 2.73TiB used 2.32TiB path /dev/sda > devid 5 size 2.73TiB used 2.32TiB path /dev/sdb > devid 6 size 1.82TiB used 1.41TiB path /dev/sdc > > Btrfs v3.12 > Data, RAID0: total=1.86TiB, used=1.85TiB > Data, single: total=5.65TiB, used=5.64TiB > System, RAID1: total=32.00MiB, used=732.00KiB > Metadata, RAID1: total=10.00GiB, used=8.69GiB > > btrfs device delete /dev/sdf /mnt/Store > ERROR: error removing the device '/dev/sdf' - Input/output error Seems it can only be removed if all of the data on that device are successfully migrated to other devices. > > I've tried rebalancing as much of the data off the drive I can. But > there's still bits in that 77GB that's good data. > > Is there a way of having btrfs skip around the input/output error and > then force the drive to remove? It's a valid question for both raid0 and single data profiles, if there will one day be a possibility to tolerate read errors, migrate what can be migrated and then permit (bad) device removal. Already a scrub would identify corrupt files. An additional feature would be a way to cause corrupted files to be easily deleted. In a case of multiple device raid0, without a regular balance being a requirement, I could very easily start with two disks, add two more disks, and so on, and end up with a significant amount of completely valid data that survives a one disk failure. Clearly the file system itself is OK due to metadata raid1. Chris Murphy-- To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html