On Thu, Jul 28, 2011 at 04:28:04PM -0400, Chris Mason wrote:

The slow performance is probably coming from reading in the metadata
associated with the snapshot extents.  The new readahead extentions from
Arne should help once we've adapted them to it.  The easiest way to make
sure is to hit sysrq-w a few times while it is slow (or run a patched
latencytop).

We shut down the rsync backups for now, and are waiting for the
snapshots to complete deletion so we can unmount the filesystem to
reboot (missing kernel options for latencytop).  In the mean time I hit
SysRq-W quite a number of times, and nothing is showing up in kernel
messages.  However, btrfs-cleaner and/or btrfs-transaction (sometimes
both, sometimes just one) have been running high in top, taking up all
the CPU time.

-- 
Bruce Guenter [off-list ref]                http://untroubled.org/