[plug] linking to different drives

[Kenworthy Family]

and when your machine crashes like mine has randomly starting to do (hardware or
dirty power? - dont know yet) you can potentially lose much more data per corrupted
inode - though the potential time savings on fsck are definitely attractive when you
have 4 crashes in three one night.

BillK


Nick Bannon wrote:

> On Thu, Oct 21, 1999 at 02:24:53PM +0800, Matt Kemner wrote:
> > On Thu, 21 Oct 1999, Kenworthy Family wrote:
> >
> > > As an aside, it was a 17gig disk and I know the figures are rubbery and linux
> > > also reserves a buffer out of the size, but 500meg out of 4.3 gigs and
> [...]
> > It is reserved for root - so when the drive gets close to running out,
> > users can no longer write to the drive, but root can still log in and fix
> > the problem.
> > You can change the amount allocated with tune2fs
> [...]
>
> That's definitely the most important one.
>
> I've also been playing with some large partitions recently, though, and
> there's a couple of other things you can fiddle with. Using fewer inodes
> and sparse superblocks (these both have their drawbacks) I freed up a
> (little) bit of extra space.
>
> However, the big benefit was time - large ext2fs partitions take a long
> time to mount and fsck. I heard tell that a large block size might help
> and it definitely did. Going from the default 1024 byte block size to
> 4096 bytes cut my mount time from 24 seconds to 2.4 seconds, and my fsck
> time from 102 seconds to 28 seconds!
>
> mkfs.ext2 -s 1 -i 8192 -m 2 -b 4096 /dev/sda3
>                             ^^^^^^^
>
> The drawback is that if you have lots of small files or directories,
> more space is wasted on the filesystem. I'm using that partition as spool
> space for half a dozen huge dump files before they get put onto tape,
> so it's insignificant.
>
> Nick.
>
> --
>   Nick Bannon  | "I made this letter longer than usual because
> nick at it.net.au | I lack the time to make it shorter." - Pascal