On Sat, Sep 26, 2009 at 8:19 AM, Greg Smith <gsmith@gregsmith.com> wrote:
>
> Another problem spot are checkpoints. If you dirty a very large buffer
> cache, that whole thing will have to get dumped to disk eventually, and on
> some workloads people have found they have to reduce shared_buffers
> specifically to keep this from being too painful.
Hi Greg,
Is this the case even if checkpoint_completion_target is set close to 1.0?
If you dirty buffers fast enough to dirty most of a huge
shared_buffers area between checkpoints, then it seems like lowering
the shared_buffers wouldn't reduce the amount of I/O needed, it would
just shift the I/O from checkpoints to the backends themselves.
It looks like checkpoint_completion_target was introduced in 8.3.0
Cheers,
Jeff