Help backing up a computer

[nvelissa]

I am new to Linux and am having trouble finding resources for what I am trying to do that I can understand. For my school's neuroimaging research lab we have our own network of 5 machines. "Snoopy2" is our host/server and "Homer" is the machine that we want to back up to. We want our data on "Snoopy2" to backup weekly to "Homer". We are not interested in passwords, secure files, or anything like that I have seen. We just want a backup of our medical data in case something happens to our main machine, so nothing too fancy.

Like I said, I have to be able to do this however, I have not been able to find someone who knows this and can explain it to a new linux user. If you could help and dummy it down for me with the simplest possible way to make this happen, my lab would greatly appreciate it.

[Touchtecservers]

Firstly, how is this data stored? Is it some sort of database system, or is it just a bunch of files in folders?

Also, what operating systems do these machines run? This information would be useful to decide what would be the best advice to give.

[elija]

Have a look at Back in Time. You can do scheduled backups from an hour to weekly.

Although it appears to take a copy every time, unless the file has actually changed all you get is a hard link. So you could mount a folder on homer onto snoopy2 and backup to that every hour.

[nvelissa]

we want to backup one folder with lots of subfolders and files as well as two programs "BRAINS2" and "AFNI"

snoopy2 is running fedora core 4 and homer is running on fedora core 6

[Touchtecservers]

Then you could do anything from setting up a cron job to run every week or something...

tar -zcf /path/to/save/file.tgz /path/to/backup1 /path/to/backup2/ ...

[nvelissa]

Then you could do anything from setting up a cron job to run every week or something...

tar -zcf /path/to/save/file.tgz /path/to/backup1 /path/to/backup2/ ...

I am sorry, but like I said I am brand new and do not know that much about this stuff. Can you tell me how to do all that and be as moron-proof and simple as possible?

[Touchtecservers]

No, it's my apology for being lazy and not explaining what I wrote, so sorry

If you know where you want to save the file, and where the files are that you want to back up, then the command itself should not be that difficult to explain.

tar -zcf:

This means create a zlib (gz) compressed file of the name that you choose. Then the rest of the command specifies what directories or files to back up.

Lets say that you had your machines accessible through the /mnt directory:
Then, on either machine (I suggest running it from the backup side):
--------------------- SNIP HERE---------
#!/bin/sh

tar -zcf /mnt/homer/backup`date +%d-%m-%y`.tgz /mnt/snoopy2/path1 /mnt/mnt/snoopy2/path2

--------------------- SNIP HERE---------
What the "backtick" date... "backtick" do is save the file as backup-24-04-2009.tgz (if it was run today) making sure that the backups aren't overwritten each week.

Now to get this to run weekly
Most distributions of linux make it nice and easy, by providing a cron.weekly folder in the /etc directory.. I'm going to assume that yours does too.

Just copy the code above (after changing the paths) into your /etc/cron.weekly/ folder. Make sure you call the script something rememberable (like backup.sh). Then on the command line:

chmod +x /etc/cron.weekly/backup.sh

What this command does is set the file so that it can run (otherwise it will not be able to be executed).

Im probably a bit out with a few of the things i said here tho, and ur best option (for a proper backup) is to use some other solution.

[jmadero]

check out sbackup, you can do networked back ups, easy GUI