From owner-freebsd-current Sat Nov 17 11:23:33 2001 Delivered-To: freebsd-current@freebsd.org Received: from mail.switch.aust.com (ppp63.adsl238.pacific.net.au [203.143.238.63]) by hub.freebsd.org (Postfix) with ESMTP id 837EB37B41F for ; Sat, 17 Nov 2001 11:23:19 -0800 (PST) Received: from smtp2.switch.aust.com (IDENT:root@ras.wa [10.0.1.140]) by mail.switch.aust.com (8.9.3/8.9.3) with ESMTP id FAA31017; Sun, 18 Nov 2001 05:22:48 +1000 X-Authentication-Warning: server1.ussbris: Host IDENT:root@ras.wa [10.0.1.140] claimed to be smtp2.switch.aust.com Received: from freebsd06.udt (nelsont@freebsd06.udt [10.0.2.56]) by smtp2.switch.aust.com (8.11.3/8.11.3) with SMTP id fAHJgrt23040; Sun, 18 Nov 2001 03:42:53 +0800 Received: by freebsd06.udt (sSMTP sendmail emulation); Sun, 18 Nov 2001 03:18:15 +0800 Date: Sun, 18 Nov 2001 03:18:15 +0800 From: Trent Nelson To: freebsd-current@FreeBSD.org Cc: matusita@jp.freebsd.org, jkh@winston.freebsd.org, hiroo@oikumene.gcd.org Subject: Sysinstall is still horribly broken. Message-ID: <20011118031815.A17408@freebsd06.udt> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline User-Agent: Mutt/1.2.5i Sender: owner-freebsd-current@FreeBSD.ORG Precedence: bulk List-ID: List-Archive: (Web Archive) List-Help: (List Instructions) List-Subscribe: List-Unsubscribe: X-Loop: FreeBSD.ORG I've just finished a three-day download, at a monsterous 2.0KB/sec, of ``nopkg-5.0-CURRENT-20011115-JPSNAP.iso'' from snapshots.jp.cur- rent.org. I had to toss my 20011111 snapshot because it still had the /mnt/dev sysinstall problem which was apparently fixed as of 20011112. It isn't. I'm installing on ad0s3. It has previously been partitioned, labeled and newfs'd into the following format: /dev/ad0s3a: / 128MB /dev/ad0s3b swap 128MB /dev/ad0s3e: /var 128MB /dev/ad0s3f: /var/tmp 128MB /dev/ad0s3g: /usr remainder (approx. 5GB). The filesystems were created with: newfs -b 16384 -f 2048 -U ... Sysinstall dies at the same place, everytime. Lets take the simple installation of ``bin'' as an example. 1. Boot the snapshot. Sysinstall main menu comes up. 2. -> Custom 3. -> Label 4. Label the partitions as mentioned above -- enabling softupdates. (i.e. ``ad0s3g /usr 5585MB UFS+S N'') 5. <- `Q' 6. -> Distributions 7. -> Custom 8. [x] bin 9. <- Exit 10. <- Exit 11. -> Media 12. <- CD/DVD 13. -> Commit 14. Last Chance? [Yes] [ Writing partition information to drive ad0 ] vty2> DEBUG: Scanning disk ad0 for root filesystem vty2> DEBUG: Scanning disk ad0 for swap partitions [ Warning: Using existing root partition. It will be assumed ] [ that you have the appropriate device entries already in /dev. ] -> [OK] vty2> fstab: /etc/fstab:0: No such file or directory vty2> fstab: /etc/fstab:0: No such file or directory vty2> boot_crunch: fsck_4.2bsd not compiled in vty2> usage: boot_crunch ..., where is one of: hostname pwd rm sh -sh test [ cpio dhclient fsck ifconfig mount_nfs newfs route rtsol slattach tunefs find minigzip gzip gunzip zcat sed arp pccardc pccard ppp sysinstall usbd usbdevs boot_crunch [ Warning: fsck returned status of 1 for /dev/ad0s3a. ] [ This partition may be unsafe to use. ] -> [OK] [ Unable to add /mnt/dev/ad0s3b as a swap device. No such file or ] [ directory ] -> [OK] [ Error mounting /mnt/dev/ad0s3g on /mnt/usr : No such file or ] [ directory ] vty2> fstab: /etc/fstab:0: No such file or directory vty2> fstab: /etc/fstab:0: No such file or directory vty2> fsck: cannot open `/mnt/dev/ad0s3g' : No such file or directory vty2> fstab: /etc/fstab:0: No such file or directory vty2> tunefs: /mnt/dev/ad0s3g: No such file or directory ---- This occurs for every device, as you'd expect. So, everything gets stored to /dev/ad0s3a, which is only 128MB, which'll obviously fill up pretty quickly and fail. 1. /mnt/dev is empty. Why isn't /dev being used? The correct entries reside in there. 2. It seems like /etc/fstab should exist. It doesn't, so fsck, I'm guessing, defaults to thinking this is a 4.2bsd filesystem, which the boot_crunch'ified fsck doesn't have support for. Last time I ran into this, I fluked it. I pressed Ctrl-C during the bin installation after it had put the tools like mkdir, ln et. al. in /mnt/bin. Then I simply ln'd /mnt/dev /dev, and all was well. To add salt to my wound, my dc0 isn't recognised because for it to work, I need a kernel compiled with ``PCI_ENABLE_IO_MODES''. So I can't NFS mount my workstation's drive to get access to any tools. I can't UFS mount the stable partition and use the /bin directory from that, because boot_crunch doesn't come with mount_ufs. (Or at least there's no user-interface to mount_ufs that I can see). So. As far as I can see, without fluking Ctrl-C at the right time to get enough of the bin tools already installed, there is SFA that can be done to work-around this. Bleh. Trent. -- Trent Nelson - Software Engineer - nelsont@switch.aust.com "A man with unlimited enthusiasm can achieve almost anything." --unknown To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-current" in the body of the message