Hello,
Situation: btrfs on LVM on RAID5, formatted with default options,
mounted with noatime,compress=lzo, kernel 3.18.1. While recovering RAID
after drive failure, another drive gets a couple of SATA link errors,
and it corrupts the FS:
http://pp.siedziba.pl/tmp/btrfs-corruption/kern.log.txt
After unmounting, it no longer mounts:
Jan 12 23:25:46 dev1 kernel: [462597.674793] BTRFS info (device dm-3):
disk space caching is enabled
Jan 12 23:25:55 dev1 kernel: [462606.331759] BTRFS critical (device
dm-3): corrupt leaf, bad key order: block=9090221506560,root=1, slot=0
Jan 12 23:25:55 dev1 kernel: [462606.333262] BTRFS critical (device
dm-3): corrupt leaf, bad key order: block=9090221555712,root=1, slot=0
Jan 12 23:25:55 dev1 kernel: [462606.373448] BTRFS critical (device
dm-3): corrupt leaf, bad key order: block=9090221555712,root=1, slot=0
Jan 12 23:25:55 dev1 kernel: [462606.374744] BTRFS: Failed to read block
groups: -5
Jan 12 23:25:55 dev1 kernel: [462606.396025] BTRFS: open_ctree failed
Running "btrfsck --repair" segfaults and doesn't do anything:
http://pp.siedziba.pl/tmp/btrfs-corruption/repair.txt
Adding "--init-extent-tree" makes it run for a bit longer, and then it
hits an assertion:
http://pp.siedziba.pl/tmp/btrfs-corruption/repair-init-extent-tree.txt
After that the FS is mountable, but switches to read-only:
http://pp.siedziba.pl/tmp/btrfs-corruption/mount-after-repair-init-extent-tree.txt
Information about used space is wrong:
root@dev1:~/btrfs-progs# ./btrfs fi df /mnt/tmp/
Data, single: total=9.09TiB, used=0.00B
System, DUP: total=8.00MiB, used=0.00B
System, single: total=4.00MiB, used=0.00B
Metadata, DUP: total=21.00GiB, used=560.00KiB
Metadata, single: total=8.00MiB, used=0.00B
GlobalReserve, single: total=16.00MiB, used=0.00B
root@dev1:~/btrfs-progs# ./btrfs fi show /dev/mapper/dev1-vol3snap2
Label: none uuid: d01c6984-8194-426d-b007-e20deb02d107
Total devices 1 FS bytes used 560.00KiB
devid 1 size 14.52TiB used 9.13TiB path /dev/mapper/dev1-vol3snap2
Btrfs v3.18
This volume contained backups. After replacing the hardware it will be
reformatted, but in the meantime I can experiment on it and try various
repair methods. Any ideas what else I could try? I try repairs on LVM
snapshots, so the original volume is still untouched.
--
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
