From owner-freebsd-current Thu May 9 06:14:04 1996 Return-Path: owner-current Received: (from root@localhost) by freefall.freebsd.org (8.7.3/8.7.3) id GAA02446 for current-outgoing; Thu, 9 May 1996 06:14:04 -0700 (PDT) Received: from skynet.ctr.columbia.edu (skynet.ctr.columbia.edu [128.59.64.70]) by freefall.freebsd.org (8.7.3/8.7.3) with SMTP id GAA02427 for ; Thu, 9 May 1996 06:13:59 -0700 (PDT) Received: (from wpaul@localhost) by skynet.ctr.columbia.edu (8.6.12/8.6.9) id JAA08894; Thu, 9 May 1996 09:12:58 -0400 From: Bill Paul Message-Id: <199605091312.JAA08894@skynet.ctr.columbia.edu> Subject: Re: Libc is broken! To: msmith@atrad.adelaide.edu.au (Michael Smith) Date: Thu, 9 May 1996 09:12:56 -0400 (EDT) Cc: current@freebsd.org, wollman@lcs.mit.edu In-Reply-To: <199605090639.QAA03827@genesis.atrad.adelaide.edu.au> from "Michael Smith" at May 9, 96 04:09:17 pm X-Mailer: ELM [version 2.4 PL24] Content-Type: text Sender: owner-current@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Of all the gin joints in all the world, Michael Smith had to walk into mine and say: > Bill Paul stands accused of saying: > > > > I'd like nothing better than to dive in and debug this, but I need > > to get 2.2-current running on my machine at work to do that, and I can't > > get the damn snapshot installed because of some mysterious NFS breakage. > > Hey! What processor class are you installing on? I'm seeing NFS death > on 386 and 486 system, but not on Pentiums. It's an Intel 386/33. No math co-processor. FreeBSD 2.1.0 installs just fine, but not the 2.2 SNAPS. (I repeat: both of the last two SNAPs failed.) Other details: 3c503 (8-bit) ethernet, two small IDE drives (one 40MB, the other 50MB), old Diamond Speedstar 3.01 SVGA adapter (ET4000), 8MB of RAM. > Terry has suggested that > this is a bcopy-optimisation problem (the date that this first started > being an issue is about right too); I need to get home to test this > properly though. Well, I've had this problem with both of the last SNAPs, which means the problem must have been around for at least a couple months. When did the bcopy() optimizations go in? Like I said over in the bugs mailing list, I'm getting failures in two different ways. I'm trying to set up a dataless client (/usr is mounted via NFS from the SPARCstation IPX on my desk running SunOS 4.1.3). The install is being done via FTP, which is to say that I'm sucking the distribution over from ftp.cdrom.com (once I copied the dist files to my own machine to verify that it wasn't just poor Internet performance -- it ain't that). However I'm writing part of it out to /usr via NFS: as soon as the emergency holographic shell appears, I manually mkdir /usr and then mount_nfs -P sparc:/usr /usr. NFS seems to work initially: I can ls -l the /usr filesystem prior to sysinstall scribbling on it. However, I can never complete the installation. Writing to the internal drives works fine, but after a very short time of writing to /usr over NFS, writing hangs. The machine does _NOT_ hang: I can still ping it and it responds to the keyboard. But NFS is toast: any attempt to perform an NFS operation hangs. If I go to the emergency holographic shell and type 'df', it gets wedged. How far it gets before it hangs tends to vary: I tried it about a half dozen times, and each time it was different. It was never more than a couple dozen files though. Usually it gets up to /usr/bin/cpio and then stops. Second, trying to install _FROM_ NFS also fails. Yes Joerg, I do have the paths right: sysinstall successfully mounts the filesystem and finds all the distribution file, but it hangs after while trying to read the root.flp image. By contrast to the first case, this hang happens almost immediately. Yes I'm using the NFS secure option. (And the -P flag when I mount manually.) Yes I've tried reducing the NFS block size. (This adapter works perfectly well with the full 8K blocksize in 2.1.0.) I've tried snooping with tcpdump to look for clues: as far as I can tell, everything just stops for now reason. Again, 2.1.0 can handle both of these cases just fine. *sigh* -Bill -- ============================================================================= -Bill Paul (212) 854-6020 | System Manager Work: wpaul@ctr.columbia.edu | Center for Telecommunications Research Home: wpaul@skynet.ctr.columbia.edu | Columbia University, New York City ============================================================================= License error: The license for this .sig file has expired. You must obtain a new license key before any more witty phrases will appear in this space. =============================================================================