Re: Postgres 9.1: Adding rows to table causing too much latency in other queries

Marti Raudsepp <marti(at)juffo(dot)org> writes:
> On Mon, Dec 19, 2011 at 17:30, Sushant Sinha <sushant354(at)gmail(dot)com> wrote:
>> This never happened earlier with postgres 9.0 Is there a known issue
>> with Postgres 9.1? Or how to report this problem?

> What *did* change in 9.1 is that there's new GIN cost estimation in
> the planner. If you do EXPLAIN ANALYZE for your queries, do the plans
> differ for 9.0 or 9.1?

I trolled the commit log a bit, and AFAICS the only significant GIN
changes between 9.1 and reasonably late-model 9.0 are the cost
estimation patch and this one:
http://git.postgresql.org/gitweb/?p=postgresql.git&a=commitdiff&h=73912e7fbd1b52c51d914214abbec1cda64595f2

which makes me wonder if maybe the OP has a very large fraction of empty
or null entries in his data. Previously those would have resulted in no
insertion traffic on a GIN index, but now they do.

> Another thought -- have you read about the GIN "fast updates" feature?
> This existed in 9.0 too.

Yeah, so it seems unlikely to be that, or at least not that by itself.

regards, tom lane