From owner-freebsd-questions Tue Apr 28 00:13:39 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id AAA10353 for freebsd-questions-outgoing; Tue, 28 Apr 1998 00:13:39 -0700 (PDT) (envelope-from owner-freebsd-questions@FreeBSD.ORG) Received: from kamidake.apricot.com (root@kamidake.apricot.com [170.1.160.36]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id AAA10347 for ; Tue, 28 Apr 1998 00:13:37 -0700 (PDT) (envelope-from scanner@apricot.com) Received: from ryoohki.apricot.com (root@ryoohki.apricot.com [170.1.160.34]) by kamidake.apricot.com (8.8.8/8.8.8) with ESMTP id AAA00564 for ; Tue, 28 Apr 1998 00:12:45 -0700 (PDT) Received: from ryoohki.apricot.com (scanner@localhost) by ryoohki.apricot.com (8.6.12/8.6.12) with ESMTP id XAA04139 for ; Mon, 27 Apr 1998 23:08:16 -0700 Message-Id: <199804280608.XAA04139@ryoohki.apricot.com> X-Authentication-Warning: ryoohki.apricot.com: scanner owned process doing -bs To: questions@FreeBSD.ORG Subject: Re: lockups on machine with NFS writes to itself.. In-reply-to: Your message of "Mon, 27 Apr 1998 22:01:37 PDT." <199804280501.WAA14422@implode.root.com> Reply-To: scanner@apricot.com X-URI: http://www.apricot.com/~scanner/ X-Face: 6K2.ZvQgQ.NDQLIx.1pW(xRu*">:}&PX-Ad_!!?wU7H4L"wF"0xEwYu=8Or0V+=5?-eO1XL 7-0Hom/|]B2C7Uznyol-NVnvEk:+sod^MyB4v4qVpPDemr;b@pZdRSXu.'Gm^t0?2l,j[&t.kbc[UW x6Lz^e$K$W Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Mon, 27 Apr 1998 23:08:15 -0700 From: Scanner Sender: owner-freebsd-questions@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG > >Apr 27 12:49:19 kamidake /kernel: Out of mbuf clusters - increase maxusers! > In your case, maxusers probably isn't the best solution. Instead, add > this to your kernel config file: > > options "NMBCLUSTERS=2000" Hm, after doing this and trying my test case (make an ISO image via mkisofs that will result in a 608mb image file) it still locks up. (there is a brief period of disk activity.) The out of mbufs message will not appear for a while yet. It is curious that this seems to only happen with NFS partitions mounted on the same machine. Although the machine is locked up (all my prompts have stopped responding) it still responds to pings with no delay at all. I am going to try upping the mbuf clusters again. Note, the test I am doing is to the UW seagate drive attached to the Adaptec 2940UW controller. (Although these writes work fine is I go to the same disk, except give the local path that involves no NFS for writing (but does use it for reading.) > >Did it not used to, if it fsck'd /, reboot then to do the remount? > In the 4.3BSD days, yes. 4.4BSD has the ability to re-mount the root > filesystem without rebooting. I haven't seen the problem you're describing > with the clean flag, however. When the machine reboots after this most recent lockup I will carefully watch the console to see if there are some exact messagese I can give this list to help clarify what the problem may be. Thanks for the help, --Scanner (scanner@apricot.com) To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-questions" in the body of the message