X-Original-To: pgsql-admin-postgresql.org@localhost.postgresql.org Received: from localhost (unknown [200.46.204.144]) by svr1.postgresql.org (Postfix) with ESMTP id 0867B5E3632 for ; Fri, 27 Aug 2004 18:33:31 -0300 (ADT) 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 61550-08 for ; Fri, 27 Aug 2004 21:33:24 +0000 (GMT) Received: from news.hub.org (news.hub.org [200.46.204.72]) by svr1.postgresql.org (Postfix) with ESMTP id 792DA5E3631 for ; Fri, 27 Aug 2004 18:33:13 -0300 (ADT) Received: from news.hub.org (news.hub.org [200.46.204.72]) by news.hub.org (8.12.9/8.12.9) with ESMTP id i7RLWSLq066732 for ; Fri, 27 Aug 2004 21:32:48 GMT (envelope-from news@news.hub.org) Received: (from news@localhost) by news.hub.org (8.12.9/8.12.9/Submit) id i7RLPu7P065218 for pgsql-admin@postgresql.org; Fri, 27 Aug 2004 21:25:56 GMT From: Christopher Browne X-Newsgroups: comp.databases.postgresql.admin Subject: Re: vacuum pg_attribute causes high load Date: Fri, 27 Aug 2004 17:24:54 -0400 Organization: cbbrowne Computing Inc Lines: 56 Message-ID: References: <60oekwpsbw.fsf@dev6.int.libertyrms.info> <20040827182458.78631.qmail@web61304.mail.yahoo.com> Mime-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 8bit X-Complaints-To: usenet@news.hub.org X-message-flag: Outlook is rather hackable, isn't it? X-Home-Page: http://www.cbbrowne.com/info/ X-Affero: http://svcs.affero.net/rm.php?r=cbbrowne User-Agent: Gnus/5.1006 (Gnus v5.10.6) XEmacs/21.4 (Security Through Obscurity, linux) Cancel-Lock: sha1:Ln1pG+30TrIR7fC+L3s5ddcO8fY= To: pgsql-admin@postgresql.org 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: 200408/284 X-Sequence-Number: 14679 After a long battle with technology, kristhegambler@yahoo.com (kris pal), an earthling, wrote: >   > >>>>>>>>>>>>>>>>>>>>>>>>>> >>> 1) Can you tell me if there is any other work around to vacuum the >>> pg_attribute system table ? Like doing it in parts so that it won't >> >cause a high load. > > Chris Browne wrote: >>No, you need to let the vacuum finish. >>If you stop it early, it'll just increase system load and accomplish _nothing_. > >>>>>>>>>>>>>>>>>>>>>>>>>> > But while doing vacuum the load is getting so high that the system > is almost freezing. Right. You said that the table was consuming 2GB of space. It has to read in that 2GB, and discard the bits that aren't necessary. > Does the performance of 'Vacuum' command depend on other system/ > database parameters?  The 'vacuum pg_attribute" command never > completes, its causing high load on the server. I had to wait for 15 > mins and kill it.  Because the system load was touching 10, slowing > down the system therefore not letting anyone access the server. What's happening, is, at root, that the vacuum process loads all the data in the table (2GB, you indicated) into memory, looks at it, and writes it back out. That presumably leads to about 2GB of reads and 2GB of writes, which takes a while. If you want this to go as quickly as possible, shut the database down and start the postmaster in single user mode so that there's no other activity competing for the system's resources. > Can we vacuum the "pg_attribiute" in parts? So that 'vacuum' will > complete faster and there won't be load issues. Thanks Tom Lane and > Chris Browne for your help so far. As I said, no, you need to let the vacuum finish. It's one table, and must be vacuumed in one piece. If you stop it early, you're just wasting your time. You cannot do it in pieces, and if you did, that wouldn't make it complete faster anyways. That would make it take MORE time. Do the vacuum, and hold on until it is done. Once that vacuum is done, later vacuums will take place MUCH quicker, and in that you are clearly making enormous numbers of modifications to the table, you need to vacuum the table regularly. -- select 'cbbrowne' || '@' || 'ntlug.org'; http://cbbrowne.com/info/spiritual.html Real Programmers use: "compress -d > a.out"