Horton

Re: Memory Accounting

Jeff Davis <pgsql@j-davis.com>

View thread View on postgr.es
From: Jeff Davis <pgsql@j-davis.com>
To: Tomas Vondra <tomas.vondra@2ndquadrant.com>, Melanie Plageman <melanieplageman@gmail.com>
Cc: Soumyadeep Chakraborty <sochakraborty@pivotal.io>, Robert Haas <robertmhaas@gmail.com>, Pg Hackers <pgsql-hackers@postgresql.org>, soumyadeep2007@gmail.com
Date: 2019-09-26T20:36:46Z
Lists: pgsql-hackers 
On Thu, 2019-09-26 at 21:22 +0200, Tomas Vondra wrote:
> It's worth mentioning that those bechmarks (I'm assuming we're
> talking
> about the numbers Rober shared in [1]) were done on patches that used
> the eager accounting approach (i.e. walking all parent contexts and
> updating the accounting for them).
> 
> I'm pretty sure the current "lazy accounting" patches don't have that
> issue, so unless someone objects and/or can show numbers
> demonstrating
> I'wrong I'll stick to my plan to get this committed soon.

That was my conclusion, as well.

Regards,
	Jeff Davis

Commits

  1. Change MemoryContextMemAllocated to return Size

  2. Use Size instead of int64 to track allocated memory

  3. Add transparent block-level memory accounting

  4. Change the way pre-reading in external sort's merge phase works.

  5. Improve memory management for external sorts.

  6. In array_agg(), don't create a new context for every group.