xref: /src/distrib/notes/alpha/install

	$NetBSD: install,v 1.13 1998/09/07 21:48:08 veego Exp $	

0. Introduction

    It's getting easier with every release to install NetBSD/alpha.
    If you do have problems, don't despair; most complications you
    might encounter are very easy to fix. We strongly suggest you
    join the port-alpha list (see the section on mailing lists on
    http://www.netbsd.org) and ask questions there.  Also, please
    report any problems you've encountered or solved by using the
    mailing list or by running send-pr(1) so that they can be fixed
    for the next release.
        
1. General

    The following is a walk-through of the steps you will take
    while getting NetBSD installed on your hard disk. It's divided
    into three basic components: booting NetBSD (section 2 below),
    preparing the disk (section 3 below), and loading the operating
    system files onto the disk (section 4 below).

2. Booting NetBSD

    You have two choices of how to boot your machine. If you have a
    floppy drive, you may boot from that. This is probably the simplest
    way of getting started. If you don't have a floppy drive, you will
    need to set yourself up for a boot from a file server on the
    network, which is a little more complex.

2.1 Making and Booting a Floppy

    The 3.5", 1.44 MB boot floppy image is found under the
    NetBSD/alpha _VER distribution directory in the file
    alpha/installation/floppy/floppy-144. You need to take this disk
    image and put it on a floppy disk.

    If you have a Unix system handy, you can do this with a command
    like the following:

        dd if=floppy-144 of=/dev/rfd0a bs=18k

    If the Unix system you are using is not a NetBSD system, you will
    probably need to replace `/dev/rfd0a' with the name of the floppy
    device on your particular system.

    If you have an MS-DOS or Windows system available, you can use
    the `rawrite.exe' utility to transfer the image to a floppy
    disk.  This utility is provided with the NetBSD/i386 install
    tools, under i386/installation/misc; a documentation file,
    `rawrite.doc' is available there as well.

    Once the floppy has been made, you simply need to put it in the
    drive and type

        boot dva0

    Now you may skip to section 3.

2.2 Booting over the Network

    Booting NetBSD/alpha _VER over a network requires a BOOTP or
    DHCP server, a TFTP server and an NFS server. (These are usually
    all run on the same machine.) There are three basic stages to
    the boot:

    1.The Alpha console software sends a BOOTP request to get its
    own address, the address of the TFTP server and the file to
    download. It downloads this file, which is the second stage
    bootstrap, via TFTP and then executes it.

    2.The secondary boot program resends the BOOTP request, this
    time also locating the NFS server and root path. It mounts the
    root path via NFS and reads in and transfers to the kernel: /netbsd.

    3.The kernel probes and configures the devices, and then sends
    out another BOOTP request so it can find out its address, the NFS
    server, and path. (The kernel probably should get this information
    from the console, but it currently doesn't.) It then mounts its
    root via NFS and continues.

2.2.1 Setting Up the Server

    You will need to set up your server to serve BOOTP, TFTP and NFS. 

    The NFS setup is quite simple. If you want to run a full system
    from the network, untar the NetBSD snapshot or distribution
    into a directory on your server and NFS export that directory
    to the client. (Make sure you put a kernel there as well, and
    create the device nodes in /dev with `sh ./MAKEDEV all'. Detailed
    instructions on netbooting can be found by visiting the
    Alpha platform page of www.NetBSD.org. At the time of this
    release, the URL for the netbooting instructions is:

	http://www.netbsd.org/Ports/alpha/netboot.html

    You'll want to map root to `root' (rather than the default
    `nobody') when you export your root filesystem. A typical
    /etc/exports line on a NetBSD system would be:

        /usr/export/alpha -maproot=0 myclient.mydom.com

    If you just want to get the install kernel loaded so that you
    can download sets to the local hard drive of that machine, you
    need nothing other than the install kernel in the NFS root
    directory on your server.

    For the TFTP setup, you need to copy the second stage bootstrap,
    netboot, into an appropriately named file (I use boot.netbsd.alpha)
    in the directory used by your TFTP server. If you extracted a full
    snapshot, you can get the netboot program from /usr/mdec/netboot;
    if not, you can get this from the installation/netboot directory
    where you found the alpha distribution.

    For the BOOTP server you need to specify the: 

         hardware type (Ethernet) 
         hardware address (Ethernet MAC address) 
         IP address of the client 
         subnet mask of the client 
         address of of the TFTP/NFS server 
         name of the second stage bootstrap loaded via TFTP 
         path to the root for the client (mounted via NFS) 

    Here's an example for a Unix system running bootpd: 

         myhost.mydom.com:\ 
                 :ht=ethernet:ha=0000c0391ae4:\ 
                 :ip=192.168.1.2:sm=255.255.255.0:\ 
                 :sa=192.168.1.1:bf=boot.netbsd.alpha:rp=/usr/export/alpha:

    And here's an example for a Unix system running dhcpd:

        host axp {
                hardware ethernet 0:0:c0:39:1a:e4;
                fixed-address 192.168.1.2;
                option host-name "myhost.mydom.com";
                filename "boot.netbsd.alpha";
                option root-path "/usr/export/alpha";
                option domain-name-servers 192.168.1.1;
                option broadcast-address 255.255.255.0;
                option domain-name "my.domain";
        }

2.2.2 The Alpha Console

    The only Ethernet device the console on most Alpha systems
    knows how to boot from is the onboard Ethernet interface or a
    DEC Tulip (21040, 21041, 21140) based PCI Ethernet card. Some
    older SMC 100 Mbps cards that use this chip have been known to
    work as well. Many older systems will not be able to use the
    newer 2.0 stepping of the 21140, however. If your system appears
    not to be receiving packets, this may be the problem. (You may
    or may not be able to update your firmware to fix this; see
    the alpha port pages on www.netbsd.org for more information on
    this.)

    Once you're set up, you should be able to boot with:

            boot -proto bootp ewa0

    Systems with the `old SRM' do not have a -proto option and
    use different device names.

3. Preparing the Disk

    If you're going to be running a diskless machine, the steps so
    far have prepared you to run, and you can skip to section 5
    ("Configuration") below.

    If you are going to run NetBSD from a local hard drive, however,
    this hard drive needs to be prepared. This preparation consists
    of putting a label on the disk, which includes information on
    the sizes and placement of the partition into which the disk
    is divided, putting the boot blocks on the disk, and initialising
    the filesystems on the partitions. This work is done by the
    `install' script from the boot floppy (or boot kernel, if you
    booted it via NFS with the INSTALL kernel).

3.1 Manual Install from the Shell Prompt

    The normal installation involves running the install shell script
    and interactively configuring the file systems, and then simply
    unpacking the tar files into these followed by running MAKEDEV.

    However, it is also possible to do the installation yourself
    from the shell, and in any case it is helpful to understand
    what the install script does. The procedure is:

	1. create /etc/disktab(5)
	2. run disklabel(8),
	3. run newfs(8)
	4. mount(8) the new root on /mnt
	5. cd to /usr/mdec and run ./installboot(8)

    If you are reviewing man pages on NetBSD platforms other than
    alpha, be sure when reading installboot that you read the alpha
    version by typing: "man 8 alpha/installboot".

3.2 Running Install

    When you first boot the INSTALL kernel you will be given the
    options of `install' or `shell'. Choose `install' and the
    install script will start.

    If, at any time, you have made a mistake in the install script
    and want to abort, press ^C. This will take you to a shell
    prompt.  You can then restart the install script by typing
    `/install', or halt the machine by typing `halt'.

3.3 Answering the Install Questions

    These will for the most part be fairly obvious. You may install
    on either a SCSI or an IDE disk, and you will be prompted for
    the disk to install on. The disks in your system will be numbered
    starting at xd0 (where x is an `s' for SCSI disks, `w' for IDE
    disks) based on the SCSI ID or IDE drive order; if you have
    more than one disk, watch the boot messages carefully to see
    which ones are probed as which numbers.

    Once you've selected a disk to install on, you'll be prompted
    for the geometry. This is also displayed in the boot messages,
    and you'll be given a chance to review the boot messages again
    to get the exact figures for the number of cylinders, heads
    and sectors.

    After this you must specify the size of your partitions.
    Generally you'll be giving the sizes in cylinders; the install
    program will tell you how many bytes there are in each cylinder.

    The swap partition is the second thing you specify, after the
    root partition. Regardless of the size of your disk, you'll
    want to specify a swap partition that's at least as large as
    the amount of RAM you have, and probably not less than 64 MB
    in any case.

    If you have a small disk (under 500 MB), it's probably best to
    devote all of the disk (excepting 64 MB or more for the swap)
    to the root partition.

    If you have more space, we recommend devoting at least 32 MB,
    and preferably 48 MB, to the root partition. /usr will need
    150 MB or so if you're not installing X, 200 MB or so if you
    are. A typical organization is 50 MB for root, 150-250 MB for
    swap, and the remaining space for /usr. With enough swap space
    configured, you can make /tmp a nice, fast mfs. See man 8
    mount_mfs, and note that the mfs will require swap space for
    the largest planned amount of /tmp storage. It doesn't return
    space when files are deleted, but just keeps it its own freelist
    so the swap space required is equal to the highwater mark of
    /tmp use, plus space required to back up main memory and store
    inactive images.

    Once you've specified this information, the install script will
    write the disklabel, install boot blocks to make the disk
    bootable, initialise the filesystems, and mount them all under
    /mnt. You're now ready to go on to the next step.

4. Installing NetBSD

    To install NetBSD you'll have to get access to the tar files
    that contain the operating system, and extract them to your
    disk.  You can get access to the tar files through either a
    network or from a CD-ROM.

4.1 Preparing to Install from a CD-ROM

    All you need to do is mount the CD-ROM, which will generally
    be device cd0. (The initial boot messages will tell you what
    the CD-ROM drive is probed as.) This would be done with:

        mount -r -t cd9660 /dev/cd0a /mnt2

4.2 Preparing to Install from the Network

    The first thing you need to do is configure the loopback network
    interface, which is done with the command

        ifconfig lo0 127.0.0.1

    Then you will have to configure your Ethernet card. The command

        ifconfig -l

    will give you a list of the network interfaces on your system.
    It will show you your ethernet cards first, followed by lo0
    (the loopback interface that we configured above), ppp0 (the
    PPP interface) and sl0 (the SLIP interface).

    To configure your ethernet card, type

        ifconfig <if> inet <addr> [netmask <netmask>] [media <media>]

    Where <if> is the network card (interface), <addr> is the IP
    address, the optional <netmask> parameter is the network mask,
    and the optional <media> parameter is one of:

        10base2         BNC connector, 10 Mbps
        AUI             AUI connector, 10 Mbps
        10baseT/UTP     Twisted pair connector, 10 Mbps
        100baseTX       Twisted pair connector, 100 Mbps
        100baseFX       Fibre-optic connector, 100 Mbps
        100baseT4       T4 twisted pair interface, 100 Mbps

    If the host you are getting the data files from is not on the
    local network, you will also have to configure a gateway into
    your system. Do this with

        route add default <gateway-IP-address>

    In order to save space on the install floppy, the resolver does
    not implement the DNS protocol, ignores /etc/resolv.conf and
    does only host table lookups.  You can specify all host addresses
    as IP numbers or you can enter the host names and numbers into
    /etc/hosts. For example, you can prepare a hosts table beforehand,
    and ftp(1) it down (by IP number) to /etc/hosts. This is not
    stored on the floppy but on the temporary ramdisk filesystem,
    so it must be repeated on any subsequent reboots from floppy.

    Once networking has been configured, you may mount the directory
    with the install files via NFS, or download them via FTP.

    To mount them via nfs, type

        mount -t nfs <hostname:/path/to/nfs/volume> /mnt2

    If this volume has been exported read-only, you may need the
    `-r' option to mount.

    To download the install sets with ftp, create a directory in
    which to put them and then use the ftp client to download them.
    Mirror sites are listed at: "http://www.netbsd.org/Sites/net.html".
    A typical session might be:

        mkdir /mnt/usr/release
        cd /mnt/usr/release
        ftp 204.152.184.75	 (ftp.netbsd.org when this was written)
        [all the following commmands are given to the ftp program
            after logging in]
        prompt
        cd /pub/NetBSD/NetBSD-1.3.2/alpha/binary/sets
        mget *
        bye

    Feel free, of course, to leave off the sets that you don't need
    if you don't plan to install everything.

    You are now ready to proceed to step 4.3.

4.3 Extracting the Operating System Files

    This is quite simple. Change to the root directory of your hard
    drive (which is /mnt if you've used the standard install script
    to this point) by typing

        cd /mnt

    For this and the following commands, replace `/mnt/usr/release/'
    with the path to your NFS volume or CD-ROM if that's how you
    chose to access your install files instead.

    The sets and kernel are extracted with

	cd /mnt
        for i in base kern comp etc games man misc text; do
            tar xpzf /mnt/usr/release/$i.tgz;
	done

    or (but do NOT unpack source/toolchain.tgz from / or /mnt) perhaps:

	cd /mnt
	for i in /mnt/usr/release/[a-z]*.tgz; do
	    echo $i
	    tar xpzf $i
	done

    Now make the device nodes:

	cd /mnt/dev
	sh ./MAKEDEV all

    Unmount the file systems and halt. The exact instructions to
    type here will depend on the file systems you created, but
    typically the commands are:

	cd /
	umount /mnt/usr
	umount /mnt
	sync		# not needed but traditional
	halt

    You should now be at the SRM console's >>> prompt and can reboot
    into the new configuration (possibly after an optional power cycle)
    with a command such as:

        boot dka0

    This command might be: `boot dka100' if your drive is on ID 1.
    You can usually use `show device' to see a full list of bootable
    devices in your system). Your system will come up in single-user
    mode, ready for you to configure it.

4.4 Optional Toolchain Source Module

    The source to the toolchain components is available in:

	.../alpha/source/toolchain.tgz

    This module unpacks into ./toolchain, so:

	cd /usr/local
	tar xpzf .../toolchain.tgz