On 4/2/13 11:54 AM, Robert Haas wrote:
But, having said that, I still think the best idea is what Andres proposed, which pretty much matches my own thoughts: the bgwriter needs to populate the free list, so that buffer allocations don't have to wait for linear scans of the buffer array.
I was hoping this one would make it to a full six years of being on the TODO list before it came up again, missed it by a few weeks. The funniest part is that Amit even submitted a patch on this theme a few months ago without much feedback: http://www.postgresql.org/message-id/6C0B27F7206C9E4CA54AE035729E9C382852FF97@szxeml509-mbs That stalled where a few things have, on a) needing more regression test workloads, and b) wondering just what the deal with large shared_buffers setting degrading performance was.
I saw refactoring in this area as waiting behind it being easier to experiment with adding new processes, but that barrier has fallen now. Maybe it needs a new freelist process, maybe it doesn't, today the code needed to try both is relatively cheap.
The other thing that always seemed to stop me was never having a typical Linux system big enough to hit some of these problems available all the time. What I did this week on that front was just go buy a 24 core server with 64GB of RAM that lives in my house. I just need to keep it two floors away if I want to sleep at night.
-- Greg Smith 2ndQuadrant US g...@2ndquadrant.com Baltimore, MD PostgreSQL Training, Services, and 24x7 Support www.2ndQuadrant.com -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
