| 1 |
On Feb 13, 2012 11:15 PM, "Joerg Schilling" < |
| 2 |
Joerg.Schilling@××××××××××××××××.de> wrote: |
| 3 |
> |
| 4 |
> Grant Edwards <grant.b.edwards@×××××.com> wrote: |
| 5 |
> |
| 6 |
> > On 2012-02-13, Michael Orlitzky <michael@××××××××.com> wrote: |
| 7 |
> > > On 02/13/12 05:49, Helmut Jarausch wrote: |
| 8 |
> > >> |
| 9 |
> > >> I've written a small Python program which outputs the file names in |
| 10 |
> > >> i-node order. If this is fed into tar or cpio nearly no seeks are |
| 11 |
> > >> required during copying. |
| 12 |
> > > |
| 13 |
> > > What makes you think the inodes are sequential on-disk? |
| 14 |
> > |
| 15 |
> > Even if the i-nodes are sequential on-disk, there's no reason to think |
| 16 |
> > that the data blocks associated with the inodes are in any particular |
| 17 |
> > order with respect to the i-nodes themselves. |
| 18 |
> |
| 19 |
> Correct, there is however a really fast method using "star -copy". |
| 20 |
> |
| 21 |
> This works because there are two decoupled processes, shared memory |
| 22 |
between |
| 23 |
> them and the fact that star reads names from directories in one big chunk. |
| 24 |
> |
| 25 |
|
| 26 |
Honestly, that's news to me. Which package has star? |
| 27 |
|
| 28 |
Rgds, |
Archives