2

I'm trying to copy the latest files and directories from one location to another. I'm using the find command with a reference file. Any files newer than the reference file get copied.

find  /root/test/BACKUP/   -newer /root/test/BACKUP//reference -exec cp -r  '{}' /root/test/backup_170213_0328 ';'

The issue is if the /root/test/BACKUP/ has an sub directories these get copied separately in addition to getting copied as a sub directory. This creates duplicate files in the destination.

[root@localhost S3_Backup]# tree /root/test/BACKUP/
/root/test/BACKUP/
├── Chrysanthemum.jpg
├── Desert.jpg
├── Hydrangeas.jpg
├── Jellyfish.jpg
├── level2
│   ├── Lighthouse.jpg
│   ├── Penguins.jpg
│   ├── teddy.jpg
│   └── Tulips.jpg
└── reference

1 directory, 9 files
[root@localhost S3_Backup]# tree /root/test/backup_170213_0328
/root/test/backup_170213_0328
├── BACKUP
│   ├── Chrysanthemum.jpg
│   ├── Desert.jpg
│   ├── Hydrangeas.jpg
│   ├── Jellyfish.jpg
│   ├── level2
│   │   ├── Lighthouse.jpg
│   │   ├── Penguins.jpg
│   │   ├── teddy.jpg
│   │   └── Tulips.jpg
│   └── reference
└── level2
    ├── Lighthouse.jpg
    ├── Penguins.jpg
    ├── teddy.jpg
    └── Tulips.jpg

3 directories, 13 files

1 Answer 1

1

If you're on non-embedded Linux, or more generally if your cp command is from GNU coreutils, then you can use cp -u to copy only files that are newer than the destination. If you don't already have files under /root/test/backup_170213_0328 that are more recent than the files under /root/test/BACKUP/, then this is equivalent to what you're doing now, except that files won't be copied twice.

You should also pass the -p option to cp. Most of the time, and especially when making backups, you should preserve permissions. And you should pass -d if you want to copy symbolic links as such, rather than copy their targets. -rdp can be abbreviated as -a.

find  /root/test/BACKUP/   -newer /root/test/BACKUP//reference -exec cp -a  {} /root/test/backup_170213_0328 ';'

However, it's unclear to me why you would want to copy a whole directory tree if its root has been modified. For example, creating a file directly under /root/test/BACKUP would cause the whole tree to be copied. You should probably not copy directories.

find  /root/test/BACKUP/ ! -type d -newer /root/test/BACKUP//reference -exec cp -RPp {} /root/test/backup_170213_0328 ';'

Furthermore all files are getting copied under the root directory, instead of reproducing the directory tree. You can fix that by using a tool whose main purpose is to copy directory trees, rather than a tool whose main purpose is to copy individual files. For example, with pax:

find  /root/test/BACKUP/ ! -type d -newer /root/test/BACKUP//reference -exec pax -rw -pe {} /root/test/backup_170213_0328 ';'

Or with rsync:

find  /root/test/BACKUP/ ! -type d -newer /root/test/BACKUP//reference -exec rsync -a {} /root/test/backup_170213_0328 ';'

But if you want to make incremental backups, I suspect that what you're really after is rsync's --link-dest. Point it at your previous backup, and it'll make a backup where files that already existed are hard-linked to the previous backup.

rsync -a --link-dest=/root/test/backup_170213_0228 /root/test/BACKUP /root/test/backup_170213_0328

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.