UNIX for Dummies Questions & Answers

Hi.

First post, and Linux newbie, so maybe I'm missing something obvious:

I have a Synology NAS that is run by a Linux distribution (which?). I have had an external hard drive connected to the NAS for making backups using Synology's backup application Time Backup.

Time Backup is based on rsync, so it creates incremental backups with all identical files being replaced by symbolic links.

I had to send the external hard drive in for repair, but before I did, I wanted to make a copy of all the backup data, so I wouldn't lose all my backup versions if they had to format the drive.

I found out, however, that when I tried to copy the drive, all the symbolic links were being followed, so I quickly used up all the space on the drive I wanted to copy the data to.

After a little research I found a solution with this command, run from a terminal (I SSH into the NAS using WinSCP):

Code:

cp -R -P /volumeSATA/satashare/TimeBackup /volume1/public

This effective copied all the backup data, including all the symbolic links and not the files that they pointed to, to a folder 'TimeBackup' in the directory 'public'.

Now I have my external hard disk back, and sure enough, it's been formatted, so now I need to put the backup data back on the drive.

This should be easy enough, I just need to switch the two locations from the command above around, and everything should be copied back - but no! When I run this command:

Code:

cp -R -P /volume1/public/TimeBackup /volumeSATA/satashare

the data starts copying alright, BUT the symbolic links are followed, so the disk fills up.

Am I doing something wrong? Shouldn't the "-P" parameter prevent the links from being followed??

---------- Post updated at 10:01 PM ---------- Previous update was at 05:34 PM ----------

Correction: It uses hard links, not symbolic links!

cp makes no guarantees about hard links. I would not automatically expect that to work. I gave it a try on RedHat:

Code:

$ cd /tmp
$ mkdir one
$ touch one/file1
$ ln one/file1 one/file2
$ cp -R -P one two
$ ls -li one
total 0
3507036 -rw-r--r-- 2 xxxxx unixadm 0 Feb 24 16:28 file1
3507036 -rw-r--r-- 2 xxxxx unixadm 0 Feb 24 16:28 file2
$ ls -li two
total 0
3507039 -rw-r--r-- 1 xxxxx unixadm 0 Feb 24 16:29 file1
3507038 -rw-r--r-- 1 xxxxx unixadm 0 Feb 24 16:29 file2
$

I have one file with two hard links in directory "one" and I have two files in directory "two". You can just stat a file and discover that it is a symlink. To discover duplicate hard links you need to maintain a list of every inode you encounter. And you need to detect when you cross a mounted file system too since each filesystem can have an inode numbered 3507036.

Please clarify whether "symbolic links" means "hard links" or "soft links" by posting a directory listing of two linked files with full directory details:

Code:

ls -lisad original_file
ls -lisad link_to_file

Then again after the files have been copied with your "cp" command.

They are hard links - but I managed to solve the problem myself! This did the trick:

rsync -r -H /volume1/public/TimeBackup /volumeSATA/satashare

It took a bit of time to move the data (16-18 hours), but now it's done, and my external drive is now an exact clone of the one I sent in.

Thanks for your replies, anyway!