On Sun, Sep 14, 2008 at 8:16 PM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
> BTW, one thing I noticed was that the hash index build time for the
> "wide column" case got a lot worse after applying the patch (from 56 to
> 237 sec). The reason for this turned out to be that with the smaller
> predicted index size, the code decided not to use the pre-sorting method
> that was recently added. Reducing effective_cache_size to less than the
> index size brought the time back down, to about 54 sec. So it would
> seem that effective_cache_size is too large a cutoff value. I'm
> considering changing hashbuild to switch over at shared_buffers instead
> of effective_cache_size --- any thoughts about that?
Switching to shared_buffers gets my vote, on my test table with
50,000,000 rows it takes about 8 minutes to create an index using the
default effective_cache_size. With effective_cache_size set to 6GB
(machine has 8GB) its still going an hour later.