On Feb 7 1999, Tim Perdue wrote:
>I have been looking for a good reference on tuning performance/memory
>use/etc, but have had no luck.
>
>I have a TON of free memory on my Linux Box - over 100MB. It doesn't appear
>that Postgres uses this memory when needed - instead it seems to hit the
>hard disk for each query.
>
>Do I have to specify the amount of buffering I want it to use? I see the -B
>command, but I figured that was more for inserting records.
>
>If I have a 100MB table that I want cached in memory, do I need to use 100MB
>of buffers with the -B switch??
>
Did an answer to this question appear? (I can't find one in the list
archives.)
I've got a similar sort of problem. I've got a Linux box with 256MB
RAM and a typical postgres process use less than 3MB of RAM. From
reading the manuals, it looks like setting either the -B or -S options
might help, but it's less than clear.
At the moment, I'm waiting for a COPY command to finish. I'm trying
to load a table with 2.2 million records (basically 8 int fields & 1
date) into the database, and I've been waiting 22 hours for it to
finish. This seems unreasonably slow for an otherwise idle dual
450MHz Xeon machine with fast disks (although I am running version 6.3
because it came pre-installed from RedHat).
Jeremy
