| 1 |
On Wed, 9 Jan 2013 02:47:07 +0000 (UTC) |
| 2 |
Grant Edwards <grant.b.edwards@×××××.com> wrote: |
| 3 |
|
| 4 |
> > ZFS is designed to deal with this problem by checksumming fs blocks |
| 5 |
> > continually; it does this at the filesystem level, not at the disk |
| 6 |
> > firmware level. |
| 7 |
> |
| 8 |
> I don't understand. If you're worried about media failure, what good |
| 9 |
> does checksumming at the file level do when failing media produces |
| 10 |
> seek/read errors rather than erroneous data? When the media fails, |
| 11 |
> there is no data to checksum. |
| 12 |
|
| 13 |
Not file level - it's filesystem level. It checksums filesystem blocks. |
| 14 |
|
| 15 |
And we are not talking about failing media either, we are talking about |
| 16 |
media corruption. You appear to have conflated them. |
| 17 |
|
| 18 |
The data on a medium can corrupt, and it can corrupt silently for a |
| 19 |
long time. At some point it may deteriorate to where it passes a cusp |
| 20 |
and then you will get your first visible sign - read failure. You did |
| 21 |
not see anything that happened prior as it was silent. |
| 22 |
|
| 23 |
-- |
| 24 |
Alan McKinnon |
| 25 |
alan.mckinnon@×××××.com |
Archives