On Wed, 09 Jan 2008 18:41:41 EST, Rik van Riel said:

quoted text
> I guess a third possible time (if we want to minimize the number of
> updates) would be when natural syncing of the file data to disk, by
> other things in the VM, would be about to clear the I_DIRTY_PAGES
> flag on the inode.  That way we do not need to remember any special
> "we already flushed all dirty data, but we have not updated the mtime
> and ctime yet" state.
> 
> Does this sound reasonable?

Is it possible that a *very* large file (multi-gigabyte or even bigger database,
for example) would never get out of I_DIRTY_PAGES, because there's always a
few dozen just-recently dirtied pages that haven't made it out to disk yet?

Of course, getting a *consistent* backup of a file like that is quite the
challenge already, because of the high likelyhood of the file being changed
while the backup runs - that's why big sites often do a 'quiesce/snapshot/wakeup'
on a database and then backup the snapshot...