"corrupt leaf" on kernel 3.17.2, w/ btrfs-progs 3.17

"corrupt leaf" on kernel 3.17.2, w/ btrfs-progs 3.17

[Stefan G. Weichinger]

My rootfs on /dev/sda2 (an SSD) shows this:

Nov 06 15:37:26 hiro.oops.intern kernel: BTRFS critical (device sda2):
corrupt leaf, slot offset bad: block=100235100160,root=1, slot=142

Scrub runs through fine:

# btrfs scr stat /dev/sda2
scrub status for ea95dbd1-ef4e-48a4-9732-54e6c80b31df
	scrub started at Thu Nov  6 15:35:00 2014, running for 172 seconds
	total bytes scrubbed: 68.46GiB with 0 errors

Do I have to panic? ;-)

Pls advise how to proceed.

Should I do some repair or btrfsck from a live-medium?

I found this:

http://comments.gmane.org/gmane.comp.file-systems.btrfs/34407

but haven't yet figured out what that means to me.


# btrfs version
Btrfs v3.17

# cat /proc/version
Linux version 3.17.2-gentoo (root@hiro.oops.intern) (gcc version 4.8.3
(Gentoo 4.8.3 p1.1, pie-0.5.9) ) #1 SMP Thu Nov 6 00:53:58 CET 2014


Stefan

Re: "corrupt leaf" on kernel 3.17.2, w/ btrfs-progs 3.17

[Stefan G. Weichinger]

Am 06.11.2014 um 15:45 schrieb Stefan G. Weichinger:

> Pls advise how to proceed.
> 
> Should I do some repair or btrfsck from a live-medium?

... had some hefty kernel crashes ...

Now I am on latest sysresccd.

btrfsck v 3.16 ... fails as well

Backing up.

Seems I have to recreate the btrfs (with 3.16?) and all its subvolumes
... *sigh*

Maybe the SSD dies?

S

Re: "corrupt leaf" on kernel 3.17.2, w/ btrfs-progs 3.17

[Stefan G. Weichinger]

Am 06.11.2014 um 17:11 schrieb Stefan G. Weichinger:
> Am 06.11.2014 um 15:45 schrieb Stefan G. Weichinger:
> 
>> Pls advise how to proceed.
>>
>> Should I do some repair or btrfsck from a live-medium?
> 
> ... had some hefty kernel crashes ...
> 
> Now I am on latest sysresccd.
> 
> btrfsck v 3.16 ... fails as well
> 
> Backing up.

While I am rsyncing I get this again in dmesg:

btrfs: corrupt leaf, slot offset bad: block ...
btrfs no csum found for inode ...

Can I determine which subvol this relates to?

As far as I understand it is likely the one currently synced?

Maybe I could only recreate that one subvolume and restore?

Or do I have to recreate the whole btrfs?

S

Re: "corrupt leaf" on kernel 3.17.2, w/ btrfs-progs 3.17

[Stefan G. Weichinger]

Copied a tarball of btrfs-progs v 3.17-r1 over and ran:

btrfs check --repair /dev/sda2

it tells me

incorrect offsets 12499 blabla
items overlap, can't fix
cmds-check.c:2918 fix_item_offset: Assertion `ret` failed

... and crashes

Any help appreciated!

Re: "corrupt leaf" on kernel 3.17.2, w/ btrfs-progs 3.17

[Stefan G. Weichinger]

restoring from backup now