On 26/11/2007, Johan Booysen <johan@xxxxxxxxxxxxxxxxx> wrote: > Thanks very much! > > In the meantime I've come across this, which apparently is the correct > way to use tar for incremental "backups": > > http://www.gnu.org/software/tar/manual/html_node/tar_88.html#SEC88 > > Anyt thoughts on what that page indicates? I just can't afford to screw > this one up, so it's very important that I get it right! > > Thanks. > > -----Original Message----- > From: redhat-list-bounces@xxxxxxxxxx > [mailto:redhat-list-bounces@xxxxxxxxxx] On Behalf Of Herta Van den Eynde > Sent: 25 November 2007 23:45 > To: General Red Hat Linux discussion list > Subject: Re: tar --newer-mtime > > On 25/11/2007, Johan Booysen <johan@xxxxxxxxxxxxxxxxx> wrote: > > Dear all, > > > > Apologies if this is a silly question, but it's important enough for > me > > to ask as a double-check: > > > > I want to do a once-off "incremental backup" of a server, in such a > way > > that I create a tar archive of only data files that have been created > or > > have been modified since a defined date. > > > > Will this do the trick, for example? > > > > tar -cvf incremental.tar --newer-mtime "24 Now 2007" * > > > > I.e. only tar up files that have been CREATED or MODIFIED since > > 24.11.2007... > > > > Thanks. > > > Linux doesn't record the creation date. What you can get is: > > atime = the last time the file was opened for read > mtime = the last time the file was closed after it was opened for write > ctime = the time the inode information was updated (e.g. as a result > of a chmod or chown > > --newer-mtime will only tar files whose mtime has changed. Better to > use --after-date, which will tar both files whose mtime or ctime has > changed. > > Kind regards, > > Herta Hi Johan, I've never used GNU tar, but the GNU organisation is responsible for many of the add-ons on which many linux users rely (cf. http://directory.fsf.org/GNU/), so it looks promising. If you have a test environment, it shouldn't be too hard to test. Simply - install GNU tar - create a small directory somewhere and populate it with random files. - take your first backup - open some files for write (e.g. using vi and making a few changes) - change the ownership or permissions on a few others, - take your incremental backup. (Play with 'date' if that is more convenient.) - move the original directory aside - restore your backup. - compare the restored directory with the original one . If they are identical, you're set. Kind regards, Herta -- redhat-list mailing list unsubscribe mailto:redhat-list-request@xxxxxxxxxx?subject=unsubscribe https://www.redhat.com/mailman/listinfo/redhat-list
