BackupPC-users

Re: [BackupPC-users] Block-level rsync-like hashing dd?

2011-04-10 13:25:27
Subject: Re: [BackupPC-users] Block-level rsync-like hashing dd?
From: Les Mikesell <lesmikesell AT gmail DOT com>
To: backuppc-users AT lists.sourceforge DOT net
Date: Sun, 10 Apr 2011 12:23:18 -0500
On 4/10/11 5:19 AM, hansbkk AT gmail DOT com wrote:
> On Sun, Apr 10, 2011 at 12:16 PM, Les Mikesell<lesmikesell AT gmail DOT com>  
> wrote:
>> I've never heard of raid sync affecting the original disk(s).  I've been 
>> doing
>> it for years, first with a set of firewire external drives (which also had 
>> USB
>> but it was slower), then the sata bays.  There might be problems in adding 
>> more
>> members than originally created in the set, though.  In your situation I 
>> would
> <snip>
>
>   FYI for all, not doubting Les' good experience doing this  (using
> mdraid mirroring as a user-land tool to mirror a partition to an
> external drive for offsite backup rotation):
>
> I had a pretty extensive discussion with the Linux-RAID list on
> thisand the general conclution was that the way mdraid does the
> mirroring would just add unnecessary kruft to the resulting filesystem
> and make recovery more difficult.

I'm not sure what you mean by 'recovery' here.  The disks I clone are always 
complete and unmounted momentarily as they are failed/removed.  Otherwise I 
would fall back to the previous copy that is still offsite.   The disks should 
be indistinguishable from the active ones, so any problems related to mdraid 
should apply to them as well.

> The bottom line from them was that rather than using RAID to do the
> mirroring to a removable drive, better to just use a tool like DD.
> When I expressed that I wanted a bit more assurance of a "bit-perfect"
> mirror being done, and was directed to these enhanced versions:
>
> http://dc3dd.sourceforge.net/
> http://www.forensicswiki.org/wiki/Dc3dd
>
> Of course there are many COTS "partition cloning" tools out there as
> well if that's a better fit for a given situation.

The difference in mdraid doing the mirroring and dd is that md can do it while 
the partition is mounted and active where dd and all the other partition 
cloning 
methods I know require the device to not be written to during the copy.  You 
can 
probably do it with lvm snapshots if you have a reliable space to hold the 
changes during the copy, but my initial tests with lvm seemed very slow - i.e. 
seemed to add more overhead than raid1.  Maybe that has improved since then. 
Realistically, though, the copying process keeps the disk busy and even with 
mdraid (and probably lvm) letting you leave it mounted, it won't be practical 
to 
do any backups while the copy happens.  Partclone might be the way to go if you 
don't mind stopping backuppc while it runs.  It would save a bit of time by not 
copying the free blocks.    Or if you have some intermediate disk workspaces 
you 
might do a clonezilla 'disk to image' clone where the copy is broken into files 
under 2k each.  Then you could rsync these a remote location, but you'd want 
two 
copies there so if the disaster you are trying to protect against happens 
during 
the copy you won't have corrupted the only remote instance you have. 
Clonezilla is a full linux distribution and you normally reboot into it for 
cloning, but the same components should work on the machine running backuppc 
with the archive partition unmounted.  I think it just does a partclone with 
the 
output fed through gzip and split and written to files, often over a network 
mount (samba, nfs, or sshfs).  Clonezilla is a handy tool to use in combination 
with backuppc anyway - for example to have image copies of windows boxes as a 
starting point for a backuppc restore.  It might be nice to adapt the same 
tools 
to image the archive into something you could rsync.  You probably would not 
want to compress the image chunks in this scenario since the file data will 
already be compressed and it would give rsync a better chance at finding 
matching blocks.

-- 
    Les Mikesell
     lesmikesell AT gmail DOT com


------------------------------------------------------------------------------
Xperia(TM) PLAY
It's a major breakthrough. An authentic gaming
smartphone on the nation's most reliable network.
And it wants your games.
http://p.sf.net/sfu/verizon-sfdev
_______________________________________________
BackupPC-users mailing list
BackupPC-users AT lists.sourceforge DOT net
List:    https://lists.sourceforge.net/lists/listinfo/backuppc-users
Wiki:    http://backuppc.wiki.sourceforge.net
Project: http://backuppc.sourceforge.net/