On Fri, Sep 2, 2011 at 2:36 PM, <hansbkk AT gmail DOT com> wrote:
> Here's an idea I have completely unrelated to my problem posting, looking
> for feedback.
>
> Goal: using single large HDDs as backup media rotating them offsite, in as
> simple and bullet-proof as possible a way.
>
> Strategy:
>
> two hard drives
> one with the base server OS installed, install BackupPC 3.1 via package
> management
> the other set up as one big partition (probably ext3), top-level folder
> "BackupPC_TOPDIR"
>
> under the latter: create conf and log
>
> move the contents of /etc/backuppc and /var/logs/backuppc contents to the
> above two new directories locations,
> and the contents of /var/lib/backuppc to the TOPDIR location itself, so
> everything related to BackupPC is there in one place
>
> (here's where I need to get the proper permissions, ideally as the
> specific chmod/chown/chgrp commands to use - any pointer appreciated)
>
> leaving the original directories empty as mount points
>
> use fstab to first mount the second drive at say /mnt/sdb3, then (using
> bind mounts) /mnt/sdb3/BackupPC_TOPDIR at /var/lib/backuppc
> then (more bind mounts) /var/log/backuppc at /var/lib/backuppc/log and
> /etc/backuppc at /var/lib/backuppc/conf
>
> I should be able to clone this second drive to another two, and
> transparently swap them out, effectively creating three entirely independent
> but functionally identical BackupPC servers. One is live, the second on the
> shelf on site, the third stored at a secure offsite location. Replace A with
> B, take A offsite and bring C back to sit on the shelf, next time replace B
> with C etc, rinse and repeat.
>
> Only downside is if the boss needs to restore a specific day's version of
> the corporate vision statement RIGHT NOW and that happens to be offsite, but
> that seems a small price to pay for the simplicity of the scheme.
>
> Obviously only works as long as everything fits on one disk - in my case not
> a problem.
>
> I suppose another server with the full historical set on LVM-over-RAID array
> would handle both those issues, as long as the single drive held at least
> one full snapshot, I still think the bullet-proof-ness of my scheme is
> better than an off-site-server mirrored over a WAN.
>
> Feedback on any gotchas, and the permissions details would be greatly
> appreciated. . .
There has been a vast amount of discussion on this list covering this
topic so you should probably wade through the archives.
My approach is a 3-member software RAID1 where 2 drives are always in
the server and the 3rd is a set rotated offsite. This gives you an
always-available current history plus a disaster-recovery copy and a
spare or two. My drives are 750gig (set up some time ago) and I just
recently got a laptop-size drive to work reasonably well as the
rotating member - which took some tweaking since it has 4k sectors and
the partition had to be aligned right. With this scheme you only
have to unmount momentarily while breaking the raid, but realistically
you can't do backups while the new member is syncing because the disk
is too busy. Others are doing something similar with LVM snapshots.
If you have good network bandwidth you can also simply run another
independent instance elsewhere hitting the same targets.
--
Les Mikesell
lesmikesell AT gmail DOT com
------------------------------------------------------------------------------
Special Offer -- Download ArcSight Logger for FREE!
Finally, a world-class log management solution at an even better
price-free! And you'll get a free "Love Thy Logs" t-shirt when you
download Logger. Secure your free ArcSight Logger TODAY!
http://p.sf.net/sfu/arcsisghtdev2dev
_______________________________________________
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/
|
