Re: New GUC to sample log queries - Mailing list pgsql-hackers

From Alvaro Herrera
Subject Re: New GUC to sample log queries
Date
Msg-id 20181129173515.t2rf6hjvzbw2ijra@alvherre.pgsql
Whole thread Raw
In response to Re: New GUC to sample log queries  (Adrien NAYRAT <adrien.nayrat@anayrat.info>)
Responses Re: New GUC to sample log queries
List pgsql-hackers
On 2018-Nov-29, Adrien NAYRAT wrote:

> > =# select name, short_desc, extra_desc, category from pg_settings where category like 'Reporting%When%' ;
> > ─[ RECORD 1 ]────────────────────────────────────────────────────────────────────────────────────────────────────
> > name       │ log_min_duration_statement
> > short_desc │ Sets the minimum execution time above which statements will be logged.
> > extra_desc │ Zero prints all queries. -1 turns this feature off.
> > category   │ Reporting and Logging / When to Log

> > ─[ RECORD 4 ]────────────────────────────────────────────────────────────────────────────────────────────────────
> > name       │ log_statement_sample_rate
> > short_desc │ Fraction of statements to log.
> > extra_desc │ 1.0 logs all statements.
> > category   │ Reporting and Logging / When to Log

> I like your short_desc, so I propose:
> 
> short_desc: "Fraction of statements over log_min_duration_statement to log"
> long_desc: "If we only want a sample, use a value between 0 (never log) and
> 1.0 (always log)"

Sounds good to me (we -> you, I suppose?).  I'd tweak the extra_desc for
log_min_duration_statement too, because we no longer log "all" when the
rate is <1; maybe "Zero prints all queries, subject to
log_statement_sample_rate.  -1 turns this feature off."

I just noticed we don't use "you" anywhere in the descs, except for this
one:

name       │ log_hostname
short_desc │ Logs the host name in the connection logs.
extra_desc │ By default, connection logs only show the IP address of the connecting host. If you want them to show the
hostname you can turn this on, but depending on your host name resolution setup it might impose a non-negligible
performancepenalty.
 
category   │ Reporting and Logging / What to Log

Probably not worth fussing about :-)

> If you agree with that, I will send another patch (I will complete
> postgresql.conf.sample with the same description).

+1

-- 
Álvaro Herrera                https://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services


pgsql-hackers by date:

Previous
From: Dmitry Dolgov
Date:
Subject: Re: Changing the autovacuum launcher scheduling; oldest table first algorithm
Next
From: Alvaro Herrera
Date:
Subject: Re: [PROPOSAL] extend the object names to the qualified names inpg_stat_statements