On Tue, Mar 21, 2017 at 8:18 PM, Peter Geoghegan <pg@bowt.ie> wrote:
>> Not only might that be unnecessary, but if we don't have a test
>> demonstrating the problem, we also don't have a test demonstrating
>> that a given approach fixes it.
>
> Preventing recycling from happening until we feel like it is probably
> fine. It is not fine to break it forever, though. The specific problem
> is that there is an XID stored in dead B-Tree + SP-GiST pages that is
> used in the subsequent RecentGlobalXmin interlock that determines if
> recycling is safe (if there is no possible index scan that could land
> on the dead page). You know, the _bt_page_recyclable() check.
Oh. OK, so this is not just about bloat -- it's about whether this
can be safely done at all. Somehow, I missed that.
--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company
