Re: Perfomance problem backupscript (more details and tests)
| Newsgroups | gmane.comp.archivers.star.user |
|---|---|
| Message-ID | <[email protected]> |
Hi, > > The ctime-quirk seems to be dependend on the type of > > filesystem and on the version of Linux. > > Given that ctime's purpose was to allow correct incremental backups, Was it ? Before exposing myself by refering to "old" man pages i had a look into my S.R.Bourne "Das UNIX System" of 1985 (a translation of "The UNIX System", Bell Labs, 1983). Where (in clumsy back translation) i read : "st_ctime is refreshed [if|when] the i-node is either written or changed." The books says that the filesystem was invented between 1975 and 1979. To my understanding, there were two kinds of building blocks in the database model of the early Unix filesystems: inode and data chunk. For both types there have been implemented two distinct timestamps. Straightforward. ctime is indeed the natural choice for a name change indicating timestamp. The only problem is that it always was so clearly designated to have a different meaning. > not adjusting it in rename is broken. The lack of a name change indicating timestamp is a birth defect of Unix. One should strive for a new cleaner standard but will anyway have to live with the old one for a long time. > The joys of Linux are that only popular > bugs tend to be fixed. Yeah. We poor little CD/DVD burners are nearly the last ones in the waiting line :( > Since most people don't do backups, these kinds of > things don't tend to be very popular. Hah ! That majority will get what it deserves ! >:)) > (Of those that do backups, most > don't try restoring them, so they only find out after a disaster that > things aren't right). Those ones have my full commiserations. :-| I myself was lucky never to fall victim to pseudo backups but have seen pathetic disasters happen to others. A central goal of my own efforts is to make the backup as verificable as is possible without actually restoring it. That's why ISO-9660 _is_ a valuable backup format in many usage scenarios - despite its limitations. > I guess Linux complies > with this by sometimes updating the c_time :-(. Bwahaha. (sneef) Oh yeah. Amen. Freedom's just another word for nothing left to lose. Vhat doess not kill me makes me only haerter. Have a nice day :) Thomas