Horton

Re: Cutting initdb's runtime (Perl question embedded)

Andres Freund <andres@anarazel.de>

View thread View on postgr.es
From: Andres Freund <andres@anarazel.de>
To: Dagfinn Ilmari Mannsåker <ilmari@ilmari.org>
Cc: Tom Lane <tgl@sss.pgh.pa.us>, Andreas Karlsson <andreas@proxel.se>, pgsql-hackers@postgreSQL.org
Date: 2017-04-17T16:53:15Z
Lists: pgsql-hackers 
On 2017-04-17 17:49:54 +0100, Dagfinn Ilmari Mannsåker wrote:
> Tom Lane <tgl@sss.pgh.pa.us> writes:
> 
> > There's certainly lots more that could be done in the genbki code,
> > but I think all we can justify at this stage of the development
> > cycle is to get the low-hanging fruit for testing speedups.
> 
> I threw Devel::NYTProf at it and picked some more low-hanging fruit.
> Attached are separate patches for each change, and here are the runtimes
> of genbki.pl and Gen_fmgrtab.pl, respectively, after each patch
> (averages of 5 runs, in millseconds):
> 
> master (b6dd1271): 355, 182
> 
> 1: Avoid unnecessary regex captures: 349, 183
> 2: Avoid repeated calls to SplitDataLine: 316, 158
> 3: Inline SplitDataLine: 291, 141
> 4: Inline check_natts: 287, 141
> 
> Together they shave 68ms or 19.2% off the runtime of genbki.pl and 41ms
> or 22.5% off the runtime of Gen_fmgrtab.pl

I'm a bit doubtful about improving the performance of genbki at the cost
of any sort of complication - it's only executed during the actual
build, not during initdb...  I don't see much point in doing things like
3) and 4), it's just not worth it imo.

- Andres

Commits

  1. Attempt to identify system timezone by reading /etc/localtime symlink.

  2. Make sure to run one initdb TAP test with no TZ set

  3. Use one transaction while reading postgres.bki, not one per line.

  4. Move bootstrap-time lookup of regproc OIDs into genbki.pl.