On Wed, Nov 13, 2013 at 08:10:29AM -0800, Christoph Hellwig wrote:
> On Wed, Nov 13, 2013 at 05:40:59PM +1100, Dave Chinner wrote:
> > From: Dave Chinner <dchinner@xxxxxxxxxx>
> > Large filesystems or high AG count filesystems generally have more
> > inherent parallelism in the backing storage. We shoul dmake use of
> > this by default to speed up repair times. Make xfs_repair use an
> > "auto-stride" configuration on filesystems with enough AGs to be
> > considered "multidisk" configurations.
> > This difference in elaspsed time to repair a 100TB filesystem with
> > 50 million inodes in it with all metadata in flash is:
> > Time IOPS BW CPU RAM
> > vanilla: 2719s 2900 55MB/s 25% 0.95GB
> > patched: 908s varied varied varied 2.33GB
> > With the patched kernel, there were IO peaks of over 1.3GB/s during
> > AG scanning. Some phases now run at noticably different speeds
> > - phase 3 ran at ~180% CPU, 18,000 IOPS and 130MB/s,
> > - phase 4 ran at ~280% CPU, 12,000 IOPS and 100MB/s
> > - the other phases were similar to the vanilla repair.
> > Memory usage is increased because of the increased buffer cache
> > size as a result of concurrent AG scanning using it.
> Looks good as long as you stick your promise to clean up the magic
> numbers later.
Already got a prototype patch for it.