Ahem. The actual reason that tmpfs is not widely used is that it
completely turns off synchronous writes of metadata, leading to
filesystems which can't possibly survive a reboot.
Linux already has un-agressive metadata writes in ext2fs, so ext2fs is
almost as fast as tmpfs, albeit riskier (frequent minor damage on
reboot, needlessly requiring human intervention.) Someday I need to
rewrite my boot scripts so that they do "fsck -y" and save me a log of
what got clobbered. When you're rebooting 4096 machines, you can't go
fix all the filesystems by hand. Grrrr.
> Others have mentioned that a "good" use of such a system would be for
> diskless stations, so that they would not have to transfer "temporary"
> files over the 'net because they didn't have local disk storage...
Yes, in that case an automatically-resizing memory filesystem would be
quite handy. But it's easy to crash your machine that way by running
out of memory, so it's better to use (less efficient) NFS or samba
filesystems.
-- g
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/