Puppy Linux Discussion Forum

I do regular pupsave backups on various media, and I also use grsync to keep data files backed up across those drives. I'm wondering if it is feasible or wise to use grsync instead of pupsave backup to keep pupsaves up to date across media. It writes and deletes only changed files, and would seem to save time.

Is that a bad idea?

I suppose I could just try it and boot up the synced pupsave and see if it runs okay.

Hi geo_c,
I normally use PupSave Backup.

Sometimes I use the right click on a savefolder and choose "duplicate" (exact copy is created), e.g. when I want to use the same savefolder for testing stuff without contaminating the original. But to do this the savefolder must not be in use at the time.

PupSave Backup has probably some built-in mechanism, so it can be used directly from running savefolder (but I don't know how is it with Grsync).

I wonder if you will be successfull .

Hi geo_c,

Pupsave Backup has internal logic to make a backup via tar, cp and/or rsync, depending on what files/folders are to be backed up. I don't think I can give a definitive answer to your question, since backing up methods are based on individual preferences/environments. But here is some insight.

• Rsync can be incredibly fast if you are transferring to a remote computer, that is running a rsync server.

• cp is usually faster if your are backing up to storage devices locally connected to your computer

time cp test1 test2
real	0m4.073s

echo hi >> test2
time rsync -av test1 test2
real	0m22.034s

Unnecessary Geeky Stuff Below
Why is rsync slower? Well get comfortable for a strange analogy. Let's pretend we are a librarian that just received many sets of encyclopedia books. To save shipping costs, we want to send a satelite library only the encyclopedia books that are different from theirs. For each of our books, let's count how many times the word "puppy" appears.

Book A | 12 times
Book B | 14 times
Book C | 10 times
Book D | 1700 times
Book E | 20 times

Now let's call our satellite library and ask them to do the same with their books. They tell your there results over the phone, so you can do a comparison. Theirs are:

Book A | 12 times
Book B | 14 times
Book C | 10 times
Book D | 50 times
Book E | 20 times

What's this? All of their books reference the word puppy the same amount, except book D. So we assume that book D is the only one changed. We will only send book D to the other library.

What about local copies?
OK, but what if in your library you have a public shelf of encyclopedia books that look the same to the ones your just received. To find which local books are different, we would also have to read through all of the public shelf books, finding the wor "puppy". Then we could determine which books are different on the public shelf to replace them. Instead of reading through each book, it would be much faster to just replace all of the books on the public shelf.

Conclusion For Local File Backups
If you copy files locally, you read the file once and write the file once
If you rsync, you read the source file, read the destination file, copy the parts that are different and write those different parts

Side Note
If you want to know how some backups, can be almost instantaneously, then you can research snap2, DRBD, COW, etc.

@snoring_cat

Thanks, that helps a whole lot. All of my stuff is local, so copying the pupsave to a new drive is probably fastest, though with my bulk document/data files, rsync is incredibly fast, almost instantaneous if just a few files are changed in a very large data set of 20gb.

geo_c,
there is one thing that I forgot to point out. rsync can work very fast on most file copies, because some of the first checks it does is to see if the timestamps or file sizes are different. If they look to be the same, they won't do the internal scanning I mentioned earlier. However, if you have a BIG file, like a save file that has changed, then my previous post comes into play.