Re: [PATCH 4/4] xfs: force writes to delalloc regions to unwritten

["Darrick J. Wong" <darrick.wong@oracle.com>]

On Fri, May 22, 2020 at 01:31:02PM +1000, Dave Chinner wrote:
> On Thu, May 21, 2020 at 07:53:23PM -0700, Darrick J. Wong wrote:
> > From: Darrick J. Wong <darrick.wong@oracle.com>
> > 
> > When writing to a delalloc region in the data fork, commit the new
> > allocations (of the da reservation) as unwritten so that the mappings
> > are only marked written once writeback completes successfully.  This
> > fixes the problem of stale data exposure if the system goes down during
> > targeted writeback of a specific region of a file, as tested by
> > generic/042.
> > 
> > Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
> > Reviewed-by: Christoph Hellwig <hch@lst.de>
> > Reviewed-by: Brian Foster <bfoster@redhat.com>
> 
> Observation: yesterday I forced a 4kB file create workload
> to use unwritten extents by setting an extent size hint. That caused
> about 4,500 xfs-conv kworker threads to be spawned by the workload
> which had 16 userspace processes creating files...
> 
> I expect that any sort of "create lots of small files" worklaod is
> going to cause xfs-conv kworker explosions, so be prepared for users
> to start reporting kworker explosions with this in place...

Yes, I've been running this patch internally for months and /so far/ the
conv explosions haven't generated any additional support calls.

(That said, we probably ought to constrain that a bit, there's really no
point in allowing concurrency beyond some unholy mix of AG count and the
estimated iops capacity of the storage...)

--D

> Cheers,
> 
> Dave.
> -- 
> Dave Chinner
> david@fromorbit.com