Re: "no unpinned buffers available" ? why? (hstore and plperl involved)

[Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>]

Dave Cramer <pg(at)fastcrypt(dot)com> writes:
> I've also got a customer getting this error message.
> the OS is OSX 10.3 they are using plpgsql, and shared buffers is set  
> very low
> shared_buffers = 16	

Well, the answer to that is "if it hurts, don't do that".  You couldn't
expect to process more than a very small number of very simple queries
with so few buffers.  (Example: a simple INSERT involving a btree index
will require at least four concurrently pinned buffers if there's a need
for a btree page split; a join query would require at least one buffer
per table and index involved, etc.)  Hubert was using a reasonably large
number of buffers, so his case sounds more like an actual bug, but I'd
call the above just pilot error.

			regards, tom lane