Date: Tue, 20 Apr 2021 20:03:13 -0500
From: Justin Pryzby <pryzby@telsasoft.com>
To: Andres Freund <andres@anarazel.de>
Cc: pgsql-hackers@postgresql.org, Tom Lane <tgl@sss.pgh.pa.us>
Subject: Re: terminate called after throwing an instance of 'std::bad_alloc'
Message-ID: <20210421010313.GR7256@telsasoft.com>
References: <20201007230242.rnwsbnxywvthef3w@alap3.anarazel.de>
 <20201218235607.GC30237@telsasoft.com>
 <20210417021755.xdcqaqqt4y46wzt7@alap3.anarazel.de>
 <20210417024854.GL3315@telsasoft.com>
 <20210417031837.GM3315@telsasoft.com>
 <20210418001324.GP3315@telsasoft.com>
 <20210419164130.byegpfrw46mzagcu@alap3.anarazel.de>
 <20210420053826.GM7256@telsasoft.com>
 <20210420055821.GA17201@telsasoft.com>
 <20210421002056.gjd6rpe6toumiqd6@alap3.anarazel.de>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20210421002056.gjd6rpe6toumiqd6@alap3.anarazel.de>
User-Agent: Mutt/1.9.4 (2018-02-28)
Archived-At: 
 <https://www.postgresql.org/message-id/20210421010313.GR7256%40telsasoft.com>
Precedence: bulk

On Tue, Apr 20, 2021 at 05:20:56PM -0700, Andres Freund wrote:
> On 2021-04-20 00:58:21 -0500, Justin Pryzby wrote:
> > On Tue, Apr 20, 2021 at 12:38:26AM -0500, Justin Pryzby wrote:
> > > I don't know if this is related to the other issues, but this seems leaky.
> > 
> > And it explains how the context use counter can exceed its threshold.
> > 
> > create or replace function fn() returns void language plpgsql as $$ declare rec int; begin SELECT relpages INTO rec FROM pg_class LIMIT 1; end $$;
> Right - at the moment the context can only be recreated when there's no
> JITed queries ongoing. That's why I asked whether your "real" query
> contains function calls, 

The real query has a bunch of function calls, including aggregates and custom
plpgsql functions.  I reduced the whole thing to the above.

ts=# explain analyze SELECT fn() FROM generate_series(1,999);
!       71692 kB max resident size

ts=# explain analyze SELECT fn() FROM generate_series(1,9999);
!       332852 kB max resident size

Which doesn't seem to leak if I "SELECT 1 INTO rec" instead of "SELECT relpages
INTO rec".

> and whether those functions are expensive enough to be JITed.

The real query has:
| Append  (cost=44.54..192946.35 rows=79 width=741)

That's a query over a 2 day period (midnight to midnight+2), so it's not hard
for me to believe it sometimes exceeds 100k cost units (jit_inline_above_cost),
depending on where we are in that interval, and on planner statistics.  It
might've been different when I first reported the problem, too.

I'm not sure what you mean about expensive enough to be jited ?
I think our custom functions don't have a COST set.

explain analyze is showing:

JIT:
  Functions: 201

-- 
Justin