[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: ext3 data

On Wed, Mar 20, 2002 at 03:24:20PM +0100, Ralf Hildebrandt wrote:
> > My understanding of the ext3 fs is that these
> > documents should have been restored regardless of
> > there state.  The document windows popped up but all
> > changes were lost.
> This depends on the journalling mode. The default mode writes the data
> first, and after that is commited to disk, the metadata is written.

If the documented hadn't been saved (and was only in application
memory), there really is nothing that could have been done.
Journaling isn't magic you know....

If you the data was recently saved, then in the default mode, only the
metadata is saved (to avoid the lengthy fsck check), but the data
blocks are not journaled, for performance reasons.  If you want the
data blocks to be journaled as well, you can enabled data journaling,
but there will be potentially very significate performance penalty in
that case.  (If it's light duty writing, you'll be find, but data
journaling means that all data blocks have to get written to disk
twice --- once to the journal, and once to the block's final location
in the filesystem.  This halves the maximum disk bandwidth that is
availble to users of the filesystem, since the other half is taken up
by the fiulesystem's journalling activity.)

						- Ted

[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]