On Thu, Feb 25, 2010 at 12:34 AM, Gokulakannan Somasundaram <gokul007@gmail.com> wrote:
On Thu, Feb 25, 2010 at 12:28 AM, Gokulakannan Somasundaram <gokul007@gmail.com> wrote:
That doesn't work because when you split an index page any sequential scan in progress will either see the same tuples twice or will miss some tuples depending on where the new page is allocated. Vacuum has a clever trick for solving this but it doesn't work for arbitrarily many concurrent scans.
Consider how the range scans are working today, while the page split happens.
The Seq scan should follow the right sibling to do the seq scan.
Gokul.
Actually thinking about what you suggested for a while, i think it should be possible, because the Oracle Fast Full Index scan essentially scans the index like that. I will try to think a way of doing that with Lehman and Yao...