On Mon, 2005-08-29 at 13:51 -0400, Jon Lewis wrote:
> huh? From what I've read, ext3 can handle up to 4TB filesystems. What
> problems do you run into with >1TB, but <4TB? I realize, XFS being 64-bit
> means it can have filesystems of inconceivable size...but after recent
> threads on this list about how xfs_repair needs considerable RAM in order
> to repair large fs's, abusing XFS's fs size limit seems a really bad idea.
> You may end up with a damaged fs that requires 128gb of RAM to repair.
> Then what will you do with that multi-TB damaged fs?
the fsck times and memory requirements aren't just an XFS thing. They
affect everyone. It's just that there's more multi TB XFS filesystems
(It's also not just filesystems that are affected by this - think about
databases too. There's a lot more rows in a database then there are
files on a filesystem).
For the time being, at certain sizes, people just have to get used to
the idea that if you ever have to run a consistency check it's either:
- going to require a lot of memory
- not complete in reasonable time
There's a break even point for when the research and development needed
to overcome this is viable - but it's not quite yet.
Stewart Smith (stewart@xxxxxxxxxxxxxxxx)
Description: This is a digitally signed message part