On 07/27/2012 07:52 PM, luben karavelov wrote:
>
> It's good that the bug is not in HEAD. I was testing on 9.1.4.
So was I, and while I thought I'd reproduced it I now suspect I was just
seeing shared_buffers touching.
Are you able to produce a self-contained SQL test that demonstrates the
leak?
Does the test program I posted behave differently on your system?
--
Craig Ringer