Horton

Re: [Patch] Optimize dropping of relation buffers using dlist

Amit Kapila <amit.kapila16@gmail.com>

View thread View on postgr.es
From: Amit Kapila <amit.kapila16@gmail.com>
To: "k.jamison@fujitsu.com" <k.jamison@fujitsu.com>
Cc: Kyotaro Horiguchi <horikyota.ntt@gmail.com>, "tsunakawa.takay@fujitsu.com" <tsunakawa.takay@fujitsu.com>, "tgl@sss.pgh.pa.us" <tgl@sss.pgh.pa.us>, "andres@anarazel.de" <andres@anarazel.de>, "robertmhaas@gmail.com" <robertmhaas@gmail.com>, "tomas.vondra@2ndquadrant.com" <tomas.vondra@2ndquadrant.com>, "pgsql-hackers@postgresql.org" <pgsql-hackers@postgresql.org>
Date: 2020-10-12T10:49:01Z
Lists: pgsql-hackers 
On Mon, Oct 12, 2020 at 3:08 PM k.jamison@fujitsu.com
<k.jamison@fujitsu.com> wrote:
>
> Hmm. When I repeated the performance measurement for non-recovery,
> I got almost similar execution results for both master and patched.
>
> Execution Time (in seconds)
> | s_b   | master | patched | %reg   |
> |-------|--------|---------|--------|
> | 128MB | 15.265 | 14.769  | -3.36% |
> | 1GB   | 14.808 | 14.618  | -1.30% |
> | 20GB  | 24.673 | 24.425  | -1.02% |
> | 100GB | 74.298 | 74.813  | 0.69%  |
>
> That is considering that I removed the recovery-related checks in the patch and just
> executed the commands on a standalone server.
> -       if (InRecovery && reln->smgr_cached_nblocks[forknum] != InvalidBlockNumber)
> +       if (reln->smgr_cached_nblocks[forknum] != InvalidBlockNumber)
>

Why so? Have you tried to investigate? Check if it takes an optimized
path for the non-recovery case?

-- 
With Regards,
Amit Kapila.

Commits

  1. Fix size overflow in calculation introduced by commits d6ad34f3 and bea449c6.

  2. Optimize DropRelFileNodesAllBuffers() for recovery.

  3. Optimize DropRelFileNodeBuffers() for recovery.

  4. Cache smgrnblocks() results in recovery.

  5. Add a check to prevent overwriting valid data if smgrnblocks() gives a