Hi Less,


On 11/07/2011 21:58, Les Mikesell wrote:
> BTW: The same would happen with the often so proposed "take off a disk
> of your RAID1".
> The way my 'take a disk off RAID1' works is that there are 3 spare 
> disks, with at least one always offsite in the rotation and another one 
> wouldn't be brought back if there was any reason to suspect that the 
> filesystem was corrupt as copied on the most recent.
I'm aware of the rotation there- it's just the same and only a question
on levels you do it. You have three disks and swap them at some time. I
take snapshots instead. In both cases it can happen a filesystem error
gets copied over, too. Unless we don't notice within our rotation cycle
we're lost.  It's the same- just as my physical box is located in the
same building and you have the manual step :) With enough bandwidth I'd
prefer to move my physical box out of this building but it's a question
of money.
I think I might move it to the garage, though :)

>
> I don't trust anything in the same building or anything that can be 
> corrupted by a live copy.  And I don't know enough about lvm to 
> understand how you can drbd to the live partition while keeping 
> snapshots of old copies.  I wouldn't have expected that to work.  Are 
> they really layered correctly so the lvm copy-on-write business works?
>
Yes, this works absolutely fine.  I have the physical disks as LVM
"physical volumes". Based on this created a LVM "logical volume". On
both hosts- the primary (virtual machine) one and on the second physical
one with just a single disk on it. On top of these identical LVM volumes
I've set up the drbd. Every write on the primary is transferred to the
secondary. Works pretty good. Taking a snapshot of the LVM volume
doesn't affect the drbd device at all.
Now let's assume I realize a fielsystem error (or a script wiped out
everything) it will be copied over to the secondary of course. Now I'd
stop drbd on both nodes, return to previous snapshot and tell the
secondary to be now primary. Start syncing and I have my data back. The
only thing I have to evaluate is to have the proper size of the
snapshot. If it fills up, the primary isn't affected.

Greetings

christian




------------------------------------------------------------------------------
All of the data generated in your IT infrastructure is seriously valuable.
Why? It contains a definitive record of application performance, security 
threats, fraudulent activity, and more. Splunk takes this data and makes 
sense of it. IT sense. And common sense.
http://p.sf.net/sfu/splunk-d2d-c2
_______________________________________________
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/