From owner-freebsd-current@FreeBSD.ORG Fri Dec 14 12:01:47 2007 Return-Path: Delivered-To: freebsd-current@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:4f8:fff6::34]) by hub.freebsd.org (Postfix) with ESMTP id C360C16A420 for ; Fri, 14 Dec 2007 12:01:47 +0000 (UTC) (envelope-from jhary@unsane.co.uk) Received: from unsane.co.uk (unsane-pt.tunnel.tserv2.fmt.ipv6.he.net [IPv6:2001:470:1f03:1f1::2]) by mx1.freebsd.org (Postfix) with ESMTP id 54A8413C45D for ; Fri, 14 Dec 2007 12:01:47 +0000 (UTC) (envelope-from jhary@unsane.co.uk) Received: from prawn.unsane.co.uk (150.117-84-212.staticip.namesco.net [212.84.117.150]) (authenticated bits=0) by unsane.co.uk (8.14.0/8.14.0) with ESMTP id lBEC1WPC042478 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=NO) for ; Fri, 14 Dec 2007 12:01:33 GMT (envelope-from jhary@unsane.co.uk) Message-ID: <476270A6.4050805@unsane.co.uk> Date: Fri, 14 Dec 2007 12:01:42 +0000 From: Vince User-Agent: Thunderbird 2.0.0.9 (X11/20071116) MIME-Version: 1.0 To: freebsd-current@freebsd.org X-Enigmail-Version: 0.95.2 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Subject: Little bit of a gotcha for zfs with gjournal. X-BeenThere: freebsd-current@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Discussions about the use of FreeBSD-current List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Fri, 14 Dec 2007 12:01:47 -0000 This is more for the archives than anything else. I have my home server build as a gmirror / and zfs for the rest, I was having regular kmem_map too small panics when building port even after putting in the various zfs tuning suggestions. I was thinking I must have a hardware issue or something weirder as the issue appeared after I added WRKDIRPREFIX= pointing to a new disk that wasnt on the zfs mirror (formatted ufs2 with gjournal) after i managed to panic the box building openoffice on the zfs (hadnt managed to panic it in a few months prior to that.) The problem I was having I finally managed to track down (I believe) to the fact that the default value of kern.geom.journal.cache.limit was high enough that when I started the build of the jdk15 port I was hitting the kmem limit by the combined efforts of zfs and gjournal. fixed by taking the value of kern.geom.journal.cache.limit down. Panic String: kmem_malloc(131072): kmem_map too small: 388349952 total allocated Like I said not a bug, more a tuning hint. Hope someone finds this useful although I cant see it being a common configuration. Vince