Re: btrfs fails to mount after power outage

[Qu Wenruo <quwenruo.btrfs@gmx.com>]

[-- Attachment #1.1: Type: text/plain, Size: 2222 bytes --]



On 2018年04月17日 00:07, Tom Vincent wrote:
> On 12 April 2018 at 00:25, Qu Wenruo <quwenruo.btrfs@gmx.com> wrote:
>> I'm curious about what's the underlying disk?
> 
> It's an Samsung PM951 NVMe SSD.
> 
>> Is it plain physical device? Or have other layers like bcache/lvm?
> 
> btrfs on LUKS
> 
>>> btrfs check
>> Full output please.
> 
> https://gist.githubusercontent.com/tlvince/acf51b37622c216e1c33cdc3dfbd321f/raw/d0237948bbffacd4bb8d53fdfa5f23391416c1e2/btrfs-check.txt

Unfortunately, not only extent tree, but also fs trees got corrupted:
root 259 inode 19916 errors 2000, link count wrong
	unresolved ref dir 16196710 index 2 namelen 12 name foo.gpg filetype 0
errors 3, no dir item, no dir index

Such output along other error messages means at least one tree block of
your fs trees get corrupted.

And it seems that all corrupted tree blocks belongs to subvolume 259.

>> For transid error, btrfs check --repair can fix it, but only do it when
>> that's the only problem.
> 
> I ran this (for ~12+ hours) to no avail; it appears to have been
> looping around "Btree for root 259 is fixed". I grew impatient and
> SIGINT-ed, which unsurprisingly toasted the file system once and for
> all (I rebuilt from backups at that point).

check --repair won't help much in this case.
So btrfs-restore would be your last chance to salvage data.

> 
> Full output:
> 
> https://gist.githubusercontent.com/tlvince/8060c19526aa011b0baff2b12e3873fd/raw/ecc43bd9dc7b352e490aa0bf0deac368af04e117/btrfs-check-repair.txt
> 
> Note, the system was fine for a few days after zero-log (before check
> --repair), but then hit the same transid error at boot.

Your filesystem is already *CORRUPTED*, so whatever happens is not a
surprise.

Only a filesystem which passes "btrfs check" without any problems could
be ensured to run for a long time.

Thanks,
Qu

> 
> On 13 April 2018 at 06:46, Duncan <1i5t5.duncan@cox.net> wrote:
>> What mount options?  In particular, is the discard option used (and of
>> course I'm assuming nothing as insane as nobarrier)?
> 
> noatime,compress=lzo
> 
> ... as well as some defaults: rw,noatime,compress=lzo,ssd,space_cache
> 


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 488 bytes --]