Received: from sss.sss.pgh.pa.us (sss.pgh.pa.us [206.210.65.6]) by hub.org (8.9.2/8.9.1) with ESMTP id MAA12978 for ; Sun, 14 Mar 1999 12:48:25 -0500 (EST) (envelope-from tgl@sss.pgh.pa.us) Received: from sss.sss.pgh.pa.us (localhost [127.0.0.1]) by sss.sss.pgh.pa.us (8.9.1/8.9.1) with ESMTP id MAA29695 for ; Sun, 14 Mar 1999 12:48:19 -0500 (EST) To: pgsql-sql@hub.org Subject: Re: [SQL] So what happens at 2GB? In-reply-to: Your message of Sat, 13 Mar 1999 23:00:29 -0500 (EST) <199903140400.XAA11118@hub.org> Date: Sun, 14 Mar 1999 12:48:19 -0500 Message-ID: <29692.921433699@sss.pgh.pa.us> From: Tom Lane : "Brett W. McCoy" > On Sat, 13 Mar 1999, Karl Denninger wrote: >>>>>> No. It works properly. I have had tables exceeding 10GB. >>>> >>>> How does it get past an operating system limitation on file size? >> >> There isn't one on FreeBSD, but Postgres doesn't seem to know that. >> Instead, it breaks the table up into 2G-sized file chunks. > > It may do that for performance reasons, I would guess, or for > compatibility with other operating systems that have the limitation Precisely. There was in fact a bug report against 6.4, that the feature failed to work correctly on some OS's because the split point was *exactly* 2Gb; apparently what works on these platforms is files less than 2Gb, not files less-than-or-equal-to 2Gb. 6.5 will break tables into chunks a tad smaller than 2Gb in order to avoid this problem; you might want to alter the RELSEG_SIZE constant in /users/postgres/pgsql/src/backend/storage/smgr/md.c to do the same in 6.4. regards, tom lane