Peter Eisentraut wrote:
> Am Freitag, 6. Juni 2008 schrieb Tom Lane:
>> Peter Eisentraut <peter_e@gmx.net> writes:
>>> - If we know better values, why don't we set them by default?
>> The problem is: better for what? In particular, I'm uncomfortable with
>> any changes in the direction of trying to make Postgres take over the
>> entire machine by default. I'd want some fairly explicit permission
>> from the user for that ...
>
> Yes, those are decisions we are going to have to make, eventually. But recall
> the three step process:
>
> 1. What values need changing?
shared_buffers
work_mem
maintenance_work_mem
checkpoint_segments
wal_sync_method
effective_cache_size
> I haven't seen a proposal for item 1 yet, so the rest is idle discussion at
> this time.
I think those cover the biggest low hanging fruit, async_commit is arguable.
Sincerely,
Joshua D. Drake