[PLUG] rsync

Michael M. Moore michael at writemoore.net
Tue Jun 2 19:06:28 UTC 2009


On Tue, 2009-06-02 at 08:18 -0700, Bruce KIlpatrick wrote:
> 
> The only error I see in the log file at this point is a permission 
> denied at /home/robin/.config/menus so adding an --exclude would not be 
> a problem.
> 
> After thinking about this a little more...besides wanting to "learn to 
> do this correctly", I am interested in what is really important to 
> save...email, browser, documents, pictures, etc...without having to 
> create a huge list of folders to include or exclude.  At this point it 
> is a 6 GB file that is saved, so not really a storage issue, yet.

I think the issue with backups is that there isn't necessarily a "learn
to do this correctly" that applies across the board.  It really depends
upon the purpose of the backup, on what you want it to accomplish, on
what you want to save and for how long, and on what you want to be able
to restore and under what circumstances.

I use [1] rdiff-backup, which uses (via rdiff) the rsync algorithm to
create and maintain a backup mirror of whatever file system(s) you tell
it to backup.  To quote Wikipedia, "rdiff-backup stores incremental
rdiff deltas with the backup, with which it is possible to recreate any
backup point."  That's a snippet of the technical stuff, but what I like
about it is that, most of the time, I just want to find an "oops, I
didn't mean to delete that" file.  It's easy with the backup created by
rdiff-backup, since it is just a mirror of your data, so the file is
right where it was before I deleted it.  Sometimes, I've wanted to
restore a previous version of a file that I've hopeless screwed up --
also easy enough with rdiff-backup, since all backups are time-stamped.
Only once have I done a complete restore of my whole backed-up file
system -- more complicated, but doable.  So it suits my needs really
well, but you may have different needs.

As for what to backup, since I run it from a cron every night, I have
built up a list of home sub-directories to exclude.  Basically,
everything I have no need to restore, like ~/.thumbnails,
~/local/share/Trash, and my browsers' caches.  I have only one file in
my user's home not owned by my user account: ~/.viminfo.  So I exclude
that, too, so I don't have to sudo to run the job or get the "permission
denied" error in my logs.  Since there are precious few files outside of
home I make changes to, I just copy them over to ~/Documents, so I have
a record of my changes.  (In my case, it's mostly to do with
configuration odds & ends for various programs, including apt.)  I don't
backup /etc, /usr, or /var.  If I made more extensive use of /usr/local,
I guess I would add that to my backups.  Basically, my goals are: 1)
data preservation of my own stuff, and 2) recreation of my settings and
system modifications should I ever need to reinstall, whether
"expectedly" or unexpectedly.

I take care of things like my address book, bookmarks, and my installed
software list by periodically exporting them to files in ~/Documents,
which then get included in my backups.  Gmail/IMAP preserves all the
email I want to keep and makes it readily accessible, but really
important email I also save to a file just in case Google blows up one
day.  (Works for me because I don't have all that much *really
important* email.)

But all this would be entirely inadequate for people running more
complex systems, set-ups, and software.  So it's not the correct way to
manage backups, it's just the way that works for me.  I think what you
have to decide is what do you want from your backup, then figure out
what sort of backup will be adequate to your needs.  That's about as
close as you can come to learning to do it correctly.

[1] http://rdiff-backup.nongnu.org/

-- 
Michael M.





More information about the PLUG mailing list