multi volume with tar 1534
On 9 Jun 2006 06:45:27 -0700, heavytull staggered into the Black Sun and said:
Whoa. You *never* write to the raw device node of a mounted filesystem unless you know exactly what you're doing. If you did that, yeah, you'd see completely bogus info when you tried to stat() any files on that filesystem, because writing directly to the device mangled the superblock and FAT.
Nope.
Yes, but the gain's pretty small. FAT was designed for low overhead, and if you want to compress things, you should create a filesystem anyway. This is a very good idea because then the filenames let you tell which part of which archive you've got on a disk. "archiveNNN.tar.gz.01" through "archiveNNN.tar.gz.05" etcetera, copy them to a larger disk, cat them all together, untar them. No worries. No wondering which disk is which. And you can access the files on non-Linux machines.
The nice thing about databases, you can normally always use different front-ends as long as you have a connection to the backend...
This multivolume archive thing is just too much of a PITA in the modern computing world, though, especially with floppies. CD-R* drives are cheap, USB keychain drives are cheap, and almost everything has an Ethernet card. With all that, floppies will soon only be of interest to computer archaeologists.
That's not a bad solution, but you might consider an alternative: give your client to call in...
-- Matt GThere is no Darkness in Eternity-But only Light too dim for us to see Brainbench MVP for Linux Admin mail: TRAP + SPAN don't belong ----------------------------- penguins, is Tux." --MegaHAL
How do I schedule a task to be done hourly
Linux groups from Newsgroups