[Novalug] best way to back up a home partition.

[Angelo Bertolli]

gregory pryzby wrote:
> On Thu, Sep 20, 2007 at 11:23:35AM -0400, Bonnie Dalzell wrote:
>   
>> i have two harddrives - each with an Ubuntu installation on them. they go 
>> into externally accessable drive drawers. one sits on the shelf as a 
>> backup.
>>
>> last night while trying to get alien to work (the rpm to deb program - 
>> not a game) I managed to ugrade ubuntu on the ubuntu drive in the machine. 
>> this resulted in loosing the ability to start up x - when you try startx 
>> you get the error:
>>
>> xinit connection refused (error111)
>>
>> so before doing anything else I put the backup harddrive in an booted from 
>> it and I used gnome-commander to copy the home directory from the messuped 
>> installation harddrive to a subdirectory on another harddrive.
>>
>> then it occured to me: what is the best way to do such a backup copy?
>>     
>
> Not sure what you want to do.
>
> I would rsync the directory to the backup and use rsync to 'restore'
>   

Yes, rsync is good.  But I agree with your sentiment:  first figure out 
what you want to do.

I was thinking about this the other day, and the way I see it there are 
really 3 problems I might ever be concerned with:

* User error/retrieval of old data
* Gradual failure/hard drive errors/a failure that goes unnoticed
* Catastrophic failure that stops the machine.

For the third one, a simple rsync mirror (what I do now) is sufficient, 
because the machine won't have the chance to rsync the problem also.

However, with the other two you have to be careful.  If you just simply 
have an rsync mirror and don't do real backups, AND you're doing this 
automatically, say every night... then you lose the ability to retrieve 
data that's older than the last sync, and you have the possibility of 
recording anything that has had problems.

I guess one solution to this is to always sync it by hand, and be in 
control of when that happens.  That is inconvenient, and I have a 
tendency to not sync as often as I should.  Also, it puts the burden of 
making sure the file system is intact and doesn't currently have a 
problem before creating the mirror.  Even though a human can make that 
decision, sometimes we're not always right.  And lastly, even if I know 
that all my files are fine, I don't know if another user has just 
recently messed something up in their own directory.

So the tools you use to ensure your data integrity really depend on what 
you are trying to protect against and what are acceptable losses.

Angelo

-- 
http://code.wikia.com/