Joel Fradkin wrote:
> Thank you I will look at that info.
> I did do an EXPLAIN ANALYSE on the view and could see it was doing the seq
> scan on 3 fields, so I did an index for the three fields and it then chose
> an index scan and ran in 27 seconds.
>
> I also did adjust my defaults to much smaller numbers on shared buffers (per
> the tidbits page recommendation like 8 meg for my memory size). I looked at
> http://www.desknow.com/kb/idx/0/061/article/ which recommended doing a
> vacuum verbose to determine the exact max_fsm_pages and I set the cache to
> use 25% of my available memory per the recommendation on tid bits.
Note that the effective_cache_size (if I've spelt it right) just tells
PG what your cache size is. You should set it based on what "free" tells
you about your system's use of memory.
-- Richard Huxton Archonet Ltd
