On Fri, Mar 13, 2009 at 10:33 AM, sunnyfr <[hidden email]> wrote:
> And it was obvious how during snappuller on a slave server, the query time
> was a lot longer than the rest of the time.
Did the CPU utilization drop?
It could be writing of the new files being pulled forcing parts of the
current index files out of OS cache.
iostat could also help.... look how much data is actually read from
disk for a certain number of queries with and without a snappull going