Re: log bind parameter values on error

From: Alvaro Herrera <alvherre(at)2ndquadrant(dot)com>
To: Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>
Cc: Andres Freund <andres(at)anarazel(dot)de>, Alexey Bashtanov <bashtanov(at)imap(dot)cc>, Peter Eisentraut <peter(dot)eisentraut(at)2ndquadrant(dot)com>, PostgreSQL Hackers <pgsql-hackers(at)postgresql(dot)org>
Subject: Re: log bind parameter values on error
Date: 2019-12-10 21:12:47
Message-ID: 20191210211247.GA22341@alvherre.pgsql
Views: Raw Message | Whole Thread | Download mbox | Resend email
Lists: pgsql-hackers

Alexey: I would appreciate it if you give this patch a spin. Let me
know if it does what you wanted it to do.

On 2019-Dec-10, Alvaro Herrera wrote:

> Here's a curious thing that happens with this patch. If you have
> log_duration set so that parameters are logged during the bind phase,
> and then an error occurs during the execution phase but you don't have
> log_parameters_on_error set true, the second error will log the
> parameters nonetheless ... because they were saved in the ParamListInfo
> struct by the errdetail_params() call in the check_log_durations block
> during bind.

AFAICS this has a simple solution, which is to stop saving the parameter
string in BuildParamLogString; instead, just *return* the string.
Caller can then assign it into params->paramValuesStr if appropriate
(which, in the case of errdetail_params(), it is not.) v22 does it like

There is still a much smaller issue that if you have both log_durations
set to log the params during bind, and log_parameters_on_error to true,
the parameters will appear twice. But I think that's correct by

Álvaro Herrera
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services

Attachment Content-Type Size
v22-0001-Emit-parameter-values-during-query-bind-execute-.patch text/x-diff 17.5 KB

In response to


Browse pgsql-hackers by date

  From Date Subject
Next Message Adam Lee 2019-12-10 21:34:22 Re: Memory-Bounded Hash Aggregation
Previous Message Bossart, Nathan 2019-12-10 21:11:12 pg_control_init() bug