cvs commit: src/sbin/fsck_ffs fsck.h pass5.c src/sys/ufs/ffs ffs_alloc.c ffs_softdep.c fs.h

[David Schultz]

On Mon, Feb 21, 2005, Xin LI wrote:
> 在 2005-02-20日的 18:17 -0500，David Schultz写道：
> > >   Since the summary is already re-sync'ed every 30 second, we will
> > >   not lag behind too much after a crash.  With this consideration
> > >   in mind, it is more reasonable to transfer the responsibility to
> > >   background fsck, to reduce the delay after a crash.
> > 
> > I'm not sure that I completely buy this explanation.  If an
> > application has a 1 GB temporary file open and unlinked at the
> > time of the crash, then upon reboot, this change will make it seem
> > as though I have 1 GB less space than I really do.  This could
> > lead to spurious disk full errors.  (Or will that happen anyway if
> > bgfsck hasn't recomputed all the free block bitmaps yet?)
> 
> Hmm...   Maybe we should add some constraint on this, for example, for
> volumes that fssize < 20G do the recomputation at mount time, despite
> the vfs.ffs.compute_summary_at_mount setting?  I think the situation
> only happens when bgfsck have not finished the scan yet, and on smaller
> volumes, this should not affect so much (after all, we can always set
> vfs.ffs.compute_summary_at_mount = 1 to restore the old behavior).
> 
> Should I send a HEADSUP / update UPDATING so more people will know the
> change?

No, I don't think any major warnings or notices are needed.  For
about two years, IIRC, the SoftUpdates implementation would report
spurious disk full errors when the disk was close to full, a large
file was deleted, and another large file was immediately created,
but practically nobody noticed at the time.  I think practically
nobody will notice your change either, except that they'll be glad
that the time required to boot after a crash has dropped dramatically.  ;-)

A note in the fsck manpage couldn't hurt, though.