From owner-freebsd-current Thu Mar 23 13: 4: 1 2000 Delivered-To: freebsd-current@freebsd.org Received: from apollo.backplane.com (apollo.backplane.com [216.240.41.2]) by hub.freebsd.org (Postfix) with ESMTP id 2F8E037C5BA for ; Thu, 23 Mar 2000 13:03:57 -0800 (PST) (envelope-from dillon@apollo.backplane.com) Received: (from dillon@localhost) by apollo.backplane.com (8.9.3/8.9.1) id NAA04161; Thu, 23 Mar 2000 13:03:44 -0800 (PST) (envelope-from dillon) Date: Thu, 23 Mar 2000 13:03:44 -0800 (PST) From: Matthew Dillon Message-Id: <200003232103.NAA04161@apollo.backplane.com> To: Greg Lehey Cc: Mathew Kanner , Soren Schmidt , freebsd-current@FreeBSD.ORG Subject: Re: ata + vinum problems References: <20000314182259.I17156@cs.mcgill.ca> <200003150829.JAA22516@freebsd.dk> <20000315095930.A7309@cs.mcgill.ca> <20000315162903.A10174@cs.mcgill.ca> <20000315165410.A3663@mojave.worldwide.lemis.com> Sender: owner-freebsd-current@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG Ok, this is NOT a softupdates *or* a vinum problem. This is a buffer cache problem. The problem is due to the large block size you are using when newfs'ing the filesystem coupled with problems in geteblk() which causes severe buffer cache KVM fragmentation. Softupdates exasperates the geteblk() problem. I thought I had fixed this problem but apparently not. Try applying my buf-cleanup-3.diff patch from: http://www.backplane.com/FreeBSD4/ Then go back to the configuration that was producing the problem before and see if it still occurs. Even with the above patch there is still a significant KVM fragmentation issue with large-block filesystems. You can guarentee 100% success by hacking the BKVASIZE define in sys/sys/param.h (AFTER applying the patch) from 16384 to 32768, but I wouldn't raise it much beyond that. This patch is only a partial fix to the problem. -Matt To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-current" in the body of the message