X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 05A6132AB16 for ; Thu, 7 Oct 2004 06:40:04 +0100 (BST) Received: from svr1.postgresql.org ([200.46.204.71]) by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) with ESMTP id 58139-02 for ; Thu, 7 Oct 2004 05:38:46 +0000 (GMT) Received: from svr4.postgresql.org (svr4.postgresql.org [66.98.251.159]) by svr1.postgresql.org (Postfix) with ESMTP id E1DD332A4B9 for ; Thu, 7 Oct 2004 06:38:46 +0100 (BST) Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) by svr4.postgresql.org (Postfix) with ESMTP id 519165AFE00 for ; Thu, 7 Oct 2004 04:18:33 +0000 (GMT) Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) by sss.pgh.pa.us (8.13.1/8.13.1) with ESMTP id i973CEMt023363; Wed, 6 Oct 2004 23:12:14 -0400 (EDT) To: Tatsuo Ishii Cc: matt@ymogen.net, pg@rbt.ca, awerman2@hotmail.com, scottakirkwood@gmail.com, pgsql-performance@postgresql.org Subject: Re: Caching of Queries In-reply-to: <20041007.100847.26272104.t-ishii@sra.co.jp> References: <20041004.002823.85416338.t-ishii@sra.co.jp> <012501c4aaf0$f12d0930$8300a8c0@solent> <20041007.100847.26272104.t-ishii@sra.co.jp> Comments: In-reply-to Tatsuo Ishii message dated "Thu, 07 Oct 2004 10:08:47 +0900" Date: Wed, 06 Oct 2004 23:12:14 -0400 Message-ID: <23362.1097118734@sss.pgh.pa.us> From: Tom Lane X-Virus-Scanned: by amavisd-new at hub.org X-Spam-Status: No, hits=0.0 tagged_above=0.0 required=5.0 tests= X-Spam-Level: X-Archive-Number: 200410/77 X-Sequence-Number: 8553 Tatsuo Ishii writes: > First, it's not a particular problem with pgpool. As far as I know any > connection pool solution has exactly the same problem. Second, it's > easy to fix if PostgreSQL provides a functionarity such as:"drop all > temporary tables if any". I don't like that definition exactly --- it would mean that every time we add more backend-local state, we expect client drivers to know to issue the right incantation to reset that kind of state. I'm thinking we need to invent a command like "RESET CONNECTION" that resets GUC variables, drops temp tables, forgets active NOTIFYs, and generally does whatever else needs to be done to make the session state appear virgin. When we add more such state, we can fix it inside the backend without bothering clients. I now realize that our "RESET ALL" command for GUC variables was not fully thought out. We could possibly redefine it as doing the above, but that might break some applications ... regards, tom lane