In message <[EMAIL PROTECTED]>, Jon Masters writes:
> On 2/5/06, Erez Zadok <[EMAIL PROTECTED]> wrote:
> > In message <[EMAIL PROTECTED]>, Jon Masters writes:
> 
> > > None of these seem to handle transparently concatenating files, just
> > > filesystems - so you can't create a file on the union which is larger
> > > than that supported by the underlying fs. In the specific case I'm
> > > thinking about involving Linux and MTD backed filesystems, it's a
> > > problem - something some folks I know want sorted and something I'm
> > > thinking about.
> 
> > I'm not sure that Unionfs is the right semantics to give you, say, 64-bit
> > files from individual file systems which only support 32 bits.
> > Nevertheless, our Redundant Array of File Systems (RAIF) file system does
> > support such a feature (as long as the OS natively can export a 64-bit file
> > API).
> 
> So if you have a bunch of filesystems, each of which is say 1GB and
> you need an 8GB filesystem overall, you support creating 8GB files
> automatically divided over the 8?

Yes, b/c just like in RAID, you can stripe chunks of individual files across
file systems; similarly you can merge statfs(2) contents.

> Jon.

Erez.
_______________________________________________
unionfs mailing list
[email protected]
http://www.fsl.cs.sunysb.edu/mailman/listinfo/unionfs

Reply via email to