Horton

Re: Ignore 2PC transaction GIDs in query jumbling

Michael Paquier <michael@paquier.xyz>

View thread View on postgr.es
From: Michael Paquier <michael@paquier.xyz>
To: Dagfinn Ilmari Mannsåker <ilmari@ilmari.org>
Cc: Julien Rouhaud <rjuju123@gmail.com>, Postgres hackers <pgsql-hackers@lists.postgresql.org>
Date: 2023-08-14T23:49:37Z
Lists: pgsql-hackers 
On Mon, Aug 14, 2023 at 12:11:13PM +0100, Dagfinn Ilmari Mannsåker wrote:
> As far as I could tell the only thing missing was removing
> DeallocateStmt from the list of unhandled utility statement types (and
> updating comments to match).  Updated patch attached.

Hmm.  One issue with the patch is that we finish by considering
DEALLOCATE ALL and DEALLOCATE $1 as the same things, compiling the
same query IDs.  The difference is made in the Nodes by assigning NULL
to the name but we would now ignore it.  Wouldn't it be better to add
an extra field to DeallocateStmt to track separately the named
deallocate queries and ALL in monitoring?
--
Michael

Commits

  1. Show names of DEALLOCATE as constants in pg_stat_statements

  2. Show GIDs of two-phase commit commands as constants in pg_stat_statements