zfs discuss - Jun 2010 - Apparent resilver slow down

I''ve noticed (at least on Solaris 10) that the resiver rate appears to 
slow down considerably as it nears completion.

On an eight 500G raidz2 vdev, after 28 hours zpool status reported:

             spare     DEGRADED     0     0    63
               c1t6d0  DEGRADED     0     0    11  too many errors
               c5t5d0  ONLINE       0     0     0  513G resilvered

Now after 55 hours:

             spare     DEGRADED     0     0    63
               c1t6d0  DEGRADED     0     0    11  too many errors
               c5t5d0  ONLINE       0     0     0  547G resilvered

Which looks like >500GB in the first 28 hours and 34GB in the next.

Is this an artefact of the reporting, or something else?  547G 
resilvered on a 500GB drive is a little suspicious!

-- 
Ian.