On Fri, Nov 18, 2011 at 3:18 PM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
>> So the correct number of WAL records is emitted and I see no bug there.
>
> What Thom's complaining about is that the buffer may be marked dirty
> unnecessarily, ie when there has been no actual data change.
Based upon both your feedback, I made a change to stop the block being
marked dirty, though Tom now wants that removed.
Thom, your earlier analysis showing that the md5 checksum of a
relation had changed is not happening because of the section of code
you identified. The code sets some data on the page, which would cause
the md5 checksum to change. So it cannot be the btree code at
_bt_delitems_vacuum() causing this.
--
Simon Riggs http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Training & Services