Re: Should pg 11 use a lot more memory building an spgist index?

Bruno Wolff III <bruno@wolff.to>

From: Bruno Wolff III <bruno@wolff.to>
To: Tom Lane <tgl@sss.pgh.pa.us>
Cc: pgsql-general@lists.postgresql.org
Date: 2018-10-24T08:37:48Z
Lists: pgsql-hackers
On Wed, Oct 24, 2018 at 09:33:48 +0100,
  Tom Lane <tgl@sss.pgh.pa.us> wrote:
>Bruno Wolff III <bruno@wolff.to> writes:
>> While reloading a database cluster to move from 10.5 to 11, I'm getting
>> out of memory crashes that I did see when doing reloads on pg 10.
>> The statement flagged in the log is this:
>> 2018-10-23 16:44:34.815 CDT [126839] STATEMENT:  ALTER TABLE ONLY public.iplocation
>> 	    ADD CONSTRAINT overlap EXCLUDE USING spgist (network WITH &&);
>
>Hm, there's a fair amount of new code in SP-GIST in v11, so maybe you've
>hit a memory leak in that.  Can you create a self-contained test case?

I'll try. I think I should only need the geolite data to cause the problem 
and I can share that publicly. So far the problem seems to be happening 
consistently. I'll work on this at the office, but probably won't get it 
done until the afternoon.

If I have a substantial database dump file to provide for reproducing this 
do you prefer it on a web server somewhere? I expect that mailing very 
large attachments to the lists is a bad idea.


Commits

Same data as JSON: GET /api/v1/messages/:b64id/commits the thread's linked commits as JSON, with link sources. API reference →
  1. Fix memory leak in repeated SPGIST index scans.

  2. Add support for nearest-neighbor (KNN) searches to SP-GiST