Bruce Momjian writes:
> Tom Lane wrote:
>> Another reason not to rely completely on the auto-analyze update path is
>> that it doesn't work for temp tables, since autovac can't access another
>> session's temp tables. It's also worth noting that auto-analyze will
>> never kick in at all for table
Tom Lane wrote:
> Another reason not to rely completely on the auto-analyze update path is
> that it doesn't work for temp tables, since autovac can't access another
> session's temp tables. It's also worth noting that auto-analyze will
> never kick in at all for tables of less than about 60 rows,
"Kevin Grittner" writes:
> Tom Lane wrote:
>> Another thing that struck me while looking at the code is that the
>> curpages clamp is applied to indexes too, which seems like a
>> thinko. A table occupying a few pages wouldn't likely have an
>> index as big as the table itself is.
> But not zer
Tom Lane wrote:
> Another thing that struck me while looking at the code is that the
> curpages clamp is applied to indexes too, which seems like a
> thinko. A table occupying a few pages wouldn't likely have an
> index as big as the table itself is.
But not zero pages, either.
> So what I'm
There's a thread over in pgsql-performance
http://archives.postgresql.org/pgsql-performance/2011-07/msg00046.php
in which the main conclusion was that we need to take a fresh look at the
heuristics the planner uses when dealing with small or empty relations.
The code in question is in estimate_rel_
