Dirk Wessner wrote: > thanks for this suggestion. I think in my case it would be easier to > find an exclude pattern that keeps these symlinks from being > backed up, because they can easily be rebuild. > I'm just not sure if I'm doing something wrong here or of this > is the way symlinks are treated by rsync/dirvish.
I just tested that on my UFS disk, which should do normal Unix semantics: Symlinks to files work, hardlinks to files work (same inode) but when I created a hardlink to a symlink, it just followed the link and made a hardlink to the symlinked-to file. Seems to be normal, then. Didn't try making my own C program and directly calling link(2) on the symlink, though. > To give some more details: I do a backup of student's www-dirs, > and they got their own phpMyAdmin. To save disk space, I symlinked > one phpMyAdmin installation to every student's www-home. Only > the config.inc.php is a "real" file. > I know that it should somehow be possible to use only one phpMyAdmin > installation for this, but it worked this way some time now, and > I didn't made up my mind on this. But this is really OT now. > Just to explain the many symlinks. I see your point, we did something similar for a while for TWikis. Still, maybe asking a sysadmin at a hosting company that runs MySQL (i.e., "a hosting company" ;-/) might help, that's their bread & butter after all. Yours, Bernd _______________________________________________ Dirvish mailing list [email protected] http://www.dirvish.org/mailman/listinfo/dirvish
