Date: Sun, 10 Nov 2019 22:23:52 +0100
From: Tomas Vondra <tomas.vondra@2ndquadrant.com>
To: Thomas Munro <thomas.munro@gmail.com>
Cc: James Coleman <jtc331@gmail.com>,
	PostgreSQL mailing lists <pgsql-bugs@lists.postgresql.org>
Subject: Re: BUG #16104: Invalid DSA Memory Alloc Request in Parallel Hash
Message-ID: <20191110212352.5xpzddwjs6fmuv4u@development>
References: 
 <CAAaqYe-NL4cCeaH4o121nBOWT_a5S2D-060qUqJYQp95k+wqZg@mail.gmail.com>
 <20191109182735.ewficcv2hibbyzl4@development>
 <CAAaqYe8=xUUwYp+XO6Y3YNqL0Nk+UXrqxYjT46Z+q6b-5U1N1Q@mail.gmail.com>
 <CA+hUKGJFNg0XqXiBNzOeTPsCa4tb-a00Sv5EA7jD+w26ATyUyA@mail.gmail.com>
 <CAAaqYe-VHx2O1k1x-wML3_1GHCPajugvMify=a_4cq-aHb=ZVg@mail.gmail.com>
 <CA+hUKGL0T7-Lsbp-dda5JWubAXEHMkL7j94-XW9ZhXDb4sn2+A@mail.gmail.com>
 <CAAaqYe9sMXm8e2+9n6KK2OTMdH3wBku06cpTcchu7k2wX2dAqA@mail.gmail.com>
 <20191110172925.mlkglpwzwego3wtp@development>
 <CAAaqYe_w=Q2F9tM0EyUvLn=CrwM99S3w9LE=MgghxKa6SRwdJw@mail.gmail.com>
 <CA+hUKGLyafKXBMFqZCSeYikPbdYURbwr+jP6TAy8sY-8LO0V+Q@mail.gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Disposition: inline
In-Reply-To: 
 <CA+hUKGLyafKXBMFqZCSeYikPbdYURbwr+jP6TAy8sY-8LO0V+Q@mail.gmail.com>
Precedence: bulk

On Mon, Nov 11, 2019 at 10:08:58AM +1300, Thomas Munro wrote:
>I think I see what's happening: we're running out of hash bits.
>
>> Buckets: 4194304 (originally 4194304)  Batches: 32768 (originally 4096)  Memory Usage: 344448kB
>
>Here it's using the lower 22 bits for the bucket number, and started
>out using 12 bits for the batch (!), and increased that until it got
>to 15 (!!).  After using 22 bits for the bucket, there are only 10
>bits left, so all the tuples go into the lower 1024 batches.
>

Ouch!

>I'm not sure how exactly this leads to wildly varying numbers of
>repartioning cycles (the above-quoted example did it 3 times, the
>version that crashed into MaxAllocSize did it ~10 times).
>
>Besides switching to 64 bit hashes so that we don't run out of
>information (clearly a good idea), what other options do we have?  (1)
>We could disable repartitioning (give up on work_mem) after we've run
>out of bits; this will eat more memory than it should.  (2) You could
>start stealing bucket bits; this will eat more CPU than it should,
>because you'd effectively have fewer active buckets (tuples will
>concentrated on the bits you didn't steal).

Can't we simply compute two hash values, using different seeds - one for
bucket and the other for batch? Of course, that'll be more expensive.

regards

-- 
Tomas Vondra                  http://www.2ndQuadrant.com
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services