Re: Recover array after I panicked

["Patrik Dahlström" <risca@powerlamerz.org>]

On 04/23/2017 04:48 PM, Andreas Klauer wrote:
> On Sun, Apr 23, 2017 at 10:06:15PM +0800, Brad Campbell wrote:
>> Nobody seems to have mentioned the reshape issue.
> 
> Good point.
> 
> If it was mid-reshape you need two sets of overlays, 
> create two RAIDs (one for each configuration), and 
> then find the point where it converges.
> 
>> If my reading of the code is correct (and my memory
>> is any good), simply adding a disk to a raid5 on a 
>> recent enough kernel should make the resync go backwards.
> 
> Doesn't it cut the offset by half and grow forwards...?
> 
> With growing a disk that should give you a segment where 
> data is identical for both 5-disk and 6-disk RAID-5. 
> And that's where you join them using dmsetup linear.
> 
> Before:
> 
> /dev/loop0:
>           Magic : a92b4efc
>         Version : 1.2
>     Feature Map : 0x1
>      Array UUID : 4611f41b:0464e815:8b6f9cfe:b29c56fd
>            Name : EIS:42  (local to host EIS)
>   Creation Time : Sun Apr 23 16:44:59 2017
>      Raid Level : raid5
>    Raid Devices : 5
> 
>  Avail Dev Size : 11720783024 (5588.90 GiB 6001.04 GB)
>      Array Size : 23441565696 (22355.62 GiB 24004.16 GB)
>   Used Dev Size : 11720782848 (5588.90 GiB 6001.04 GB)
>     Data Offset : 262144 sectors
>    Super Offset : 8 sectors
>    Unused Space : before=262064 sectors, after=176 sectors
>           State : clean
>     Device UUID : acd8d9fd:7b7cf9a0:f63369d1:907ffa66
> 
> Internal Bitmap : 8 sectors from superblock
>     Update Time : Sun Apr 23 16:44:59 2017
>   Bad Block Log : 512 entries available at offset 32 sectors
>        Checksum : f89bdc5 - correct
>          Events : 2
> 
>          Layout : left-symmetric
>      Chunk Size : 512K
> 
>    Device Role : Active device 0
>    Array State : AAAAA ('A' == active, '.' == missing, 'R' == replacing)
> 
> After/During grow:
> 
> /dev/loop0:
>           Magic : a92b4efc
>         Version : 1.2
>     Feature Map : 0x45
>      Array UUID : 4611f41b:0464e815:8b6f9cfe:b29c56fd
>            Name : EIS:42  (local to host EIS)
>   Creation Time : Sun Apr 23 16:44:59 2017
>      Raid Level : raid5
>    Raid Devices : 6
> 
>  Avail Dev Size : 11720783024 (5588.90 GiB 6001.04 GB)
>      Array Size : 29301957120 (27944.52 GiB 30005.20 GB)
>   Used Dev Size : 11720782848 (5588.90 GiB 6001.04 GB)
>     Data Offset : 262144 sectors
> |     New Offset : 257024 sectors
>    Super Offset : 8 sectors
>           State : clean
>     Device UUID : acd8d9fd:7b7cf9a0:f63369d1:907ffa66
> 
> Internal Bitmap : 8 sectors from superblock
> |  Reshape pos'n : 1472000 (1437.50 MiB 1507.33 MB)
> |  Delta Devices : 1 (5->6)
> 
>     Update Time : Sun Apr 23 16:45:38 2017
>   Bad Block Log : 512 entries available at offset 32 sectors
>        Checksum : fbd9a55 - correct
>          Events : 30
> 
>          Layout : left-symmetric
>      Chunk Size : 512K
> 
>    Device Role : Active device 0
>    Array State : AAAAAA ('A' == active, '.' == missing, 'R' == replacing)
> 
> Basically you have to know the New Offset 
> (search first 128M of your drives for filesystem headers, that should be it)
Let's see if I understand you correctly:

* I try to find 0x53EF (ext4 magic) within the first 128M of
/dev/sd[abcde]. Not after? This will be an indication of my "New
Offset". I need to adjust the offset a bit since the ext4 magic is
located at 0x438 offset.

> and then guess the Reshape pos'n by comparing raw data at offset X 
> (find non-zero data at identical offsets for both raid sets)

* I create a 5 and a 6 drive raid set and try to find an offset where
they both carry the same raw data. With some overlays, I should be able
to create both these raids at the same time, correct?

> 
> Regards
> Andreas Klauer
>