X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org Received: from localhost (developer.postgresql.org [64.117.224.193]) by svr1.postgresql.org (Postfix) with ESMTP id CC3E0B430CB for ; Wed, 9 Jul 2003 21:45:58 +0000 (GMT) Received: from svr1.postgresql.org ([64.117.224.193]) by localhost (svr1.postgresql.org [64.117.224.193]) (amavisd-new, port 10024) with ESMTP id 79242-10 for ; Wed, 9 Jul 2003 18:45:26 -0300 (ADT) Received: from ethereal-realms.org (thebe.ethereal-realms.org [142.179.222.205]) by svr1.postgresql.org (Postfix) with ESMTP id D03B3B439ED for ; Wed, 9 Jul 2003 18:45:25 -0300 (ADT) Received: from ethereal-realms.org (ip14.ethereal-realms.COM [192.168.1.14]) by ethereal-realms.org (8.12.6/8.12.6) with ESMTP id h69Lj8Uc012190; Wed, 9 Jul 2003 15:45:10 -0600 (MDT) Message-ID: <3F0C8CE1.3080308@ethereal-realms.org> Date: Wed, 09 Jul 2003 15:45:05 -0600 From: Martin Foster Organization: Ethereal Realms User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.4) Gecko/20030624 Netscape/7.1 (ax) X-Accept-Language: en-us, en MIME-Version: 1.0 To: Scott Marlowe , PostgreSQL Performance Subject: Re: Moving postgresql.conf tunables into 2003... References: In-Reply-To: Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit X-Virus-Scanned: by amavisd-new at postgresql.org X-Archive-Number: 200307/157 X-Sequence-Number: 2450 Scott Marlowe wrote: > > It would be nice to have a program that could run on any OS postgresql > runs on and could report on the current limits of the kernel, and make > recommendations for changes the admin might want to make. > > One could probably make a good stab at effective cache size during > install. Anything reasonably close would probably help. > > Report what % of said resources could be consumed by postgresql under > various circumstances... > One of the issues that automating the process would encounter are limits in the kernel that are too low for PostgreSQL to handle. The BSD's come to mind where they need values manually increased in the kernel before you can reach a reasonable maximum connection count. Another example is how OpenBSD will outright crash when trying to test the database during install time. It seems that most of the tests fail because the maximum amount of processes allowed is too low for the test to succeed. While FreeBSD will work just fine on those same tests. If PostgreSQL automates the configuration, that would be a plus. But also detect the platform and inform the person that these changes should be done to the kernel, sysctl or whatever in order to have that configuration run. Perl may be useful in this for a few reasons. It's portable enough to run on multiple Unix variants and the tools would be fairly standard, so the code would require less considerations for more exotic implementations. Martin Foster Creator/Designer Ethereal Realms martin@ethereal-realms.org