Using the MilaX Live CD for failsafe booting

MilaX is a small size Live CD distribution which runs completely off a CD . In addition to various other usages (see links) MilaX can also be used as failsafe image for the Solaris OS.

A failsafe image is an image for a ramdisk from which the Solaris OS can boot. This boot method is used, for example, if you do a "boot -s" in newer Solaris versions.

Scope

This article only describes how to create the image for the Solaris OS for x86 platforms.

Prerequisites

A working Solaris installation (either Solaris, Solaris Nevada, or OpenSolaris); the MilaX image for x86 described here was tested with Solaris Express Community Edition Build 111, Solaris 10 10/08 OS, and OpenSolaris 2009.06 111a.

The MilaX Live CD; for this article, I used MilaX for x86 version 0.3.3 and version 0.4.

References

Creating a failsafe image with MilaX for the Solaris OS on x86 Platforms

Preparation

To create the image with MilaX, boot from the MilaX Live CD and open a root shell.

# Note: /mnt/solaris0 is mounted to the internal hard disk in this example
#
(root@milax) # mkfile -n 500m /mnt/solaris0/milax_img

Next lofi mount the image, create a filesystem on the lofi device, and mount the lofi device:

(root@milax) # LOFI_DEV=$(  lofiadm -a /mnt/solaris0/milax_img ) && \
  newfs ${LOFI_DEV}

(root@milax) # mkdir /tmp/milax

(root@milax) # mount ${LOFI_DEV} /tmp/milax

Then copy the script copy_milax_to_img.sh (see the appendix) to the machine, make it executable using chmod, and execute it, e .g

(root@milax) # ./copy_milax_to_img.sh /tmp/milax
Starting to copy data from UFS root to /tmp/milax - this may take some time.
/tmp/milax/etc/crypto/certs/SUNWObjectCA linked to /tmp/milax/etc/certs/SUNWObjectCA
/tmp/milax/etc/rc0.d/K50pppd linked to /tmp/milax/etc/init.d/pppd
...
tar: ./svc/volatile/kcfd_door is not a file. Not dumped
tar: ./sysevent/devfsadm_event_channel/1 is not a file. Not dumped
 ###################################################################

       		MilaX now installed on /tmp/milax

 ###################################################################

(root@milax) # mv /tmp/milax/platform/i86pc/kernel/unix /tmp/milax/platform/i86pc/kernel/milax_unix

(root@milax) #  umount /tmp/milax

(root@milax) # lofiadm -d ${LOFI_DEV}

Now copy the image file, /mnt/solaris0/milax_img in this example, to a save place.

Installation

Log in as root to the target machine and copy the image to the machine; in this example, we assume the image is called milax_img and it is in the directory /milax.

bash-3.2# mkdir /tmp/milax

bash-3.2# mount $( lofiadm -a /milax/milax_img ) /tmp/milax

bash-3.2#  cp /tmp/milax/platform/i86pc/kernel/milax_unix  /milax/platform/i86pc/kernel/milax_unix

bash-3.2 # umount /tmp/milax

bash-3.2# lofiadm -d  /milax/milax_img

title MilaX 0.3.3 (32 Bit) (failsafe)
	findroot (pool_rpool,0,a)
	kernel$ /platform/i86pc/kernel/milax_unix
	module$ /milax/milax.img

That's it; now you can boot the failsafe image with MilaX for maintenance tasks on your main Solaris installation.

~~Note that the MilaX failsafe image is not machine dependent; you can copy it to any x86 machine.~~

Customizing the failsafe image

To customize the Solaris OS in the image boot into the normal Solaris OS and lofi mount the image, do the following (in this example, we assume the image is called milax_img and it is in the directory /milax).

bash-3.2# mkdir /tmp/milax

bash-3.2# mount $( lofiadm -a /milax/milax_img ) /tmp/milax

bash-3.2 # umount /tmp/milax

bash-3.2# lofiadm -d  /milax/milax_img

Troubleshooting

My changes to the failsafe image are not persistent.

To change the contents of the image, you must lofi mount the image (see above); changes made while the machine is booted from the image are not persistent.

There's no network connection after booting from the failsafe image

The Solaris OS in the failsafe image needs a DHCP server to automatically configure the network adapter. If there is no DHCP server in your environment, you must either add a static IP configuration to the image (see above) or configure the network manually every time after the image is booted.

The X Server starts automatically but not the Window manager

There's a bug in MilaX 0.4 that the Window manager does not start automatically every reboot. If this the case you only see the grey background from the X server and no menus.

To get around this bug simply kill the X Server (using CTRL-ALT-BACKSPACE) and restart the X Server using the command startx, e.g.

(alex@milx) # startx

How to use more than one MilaX failsafe image

To use more than one MilaX failsafe image use unique names for the kernel and the image, e.g.
milax040_img and milax040_unix. Note that the name of the kernel must be the same in your normal Solaris installation and in the MilaX failsafe image.

Appendix

copy_milax_to_img.sh

#!/usr/bin/ksh
#
# CDDL HEADER START
#
# The contents of this file are subject to the terms of the
# Common Development and Distribution License (the "License").
# You may not use this file except in compliance with the License.
#
# You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
# or http://www.opensolaris.org/os/licensing.
# See the License for the specific language governing permissions
# and limitations under the License.
#
# When distributing Covered Code, include this CDDL HEADER in each
# file and include the License file at usr/src/OPENSOLARIS.LICENSE.
# If applicable, add the following below this CDDL HEADER, with the
# fields enclosed by brackets "[]" replaced with your own identifying
# information: Portions Copyright [yyyy] [name of copyright owner]
#
# CDDL HEADER END
#
# Copyright 2007 Sun Microsystems, Inc.  All rights reserved.
# Use is subject to license terms.
#
# Script to install MilaX into a boot image for failsafe booting
# written by Bernd Schemmer
# based on the sript to install MilaX to ZFS
# By Alexander R. Eremin, April 4 2008.
# Thanks to Tim Foster for basic zfs-boot script
#
#
# History
#   v1.10 14.06.2009 /bs
#       added support for MilaX 0.4
#

# Make sure only root can run our script
if [ "$(id -u)" != "0" ]; then
    echo "This script must be run as root"
    exit 1
fi

# check arguments
if [ "$#" -lt 1 ] || [ $1 == "-h" ] || [ $1 == "--help" ]
then
  echo "Usage : $0 targetfilesystem"
  echo "eg. $0 /mnt/milax"
  exit 1
fi
TARGET_FS="$1"

#
# get MiIaX version
#
if [ -f /etc/release ] ; then
  MILAX_VERSION="$( head -1 /etc/release | tr -s " " | cut -f3 -d " " )"
  RELEASE_FOUND=0
else
  MILAX_VERSION="$( uname -a | tr -s "_" " " | cut -f5 -d " " )"
  RELEASE_FOUND=1
fi

case ${MILAX_VERSION} in
  0.4 )
    echo "Using MilaX version ${MILAX_VERSION}"
    ;;

  0.3.3 )
    echo "Using MilaX version ${MILAX_VERSION}"
   ;;

  * )
    if [ ${RELEASE_FOUND} = 0 ] ; then
      echo "WARNING: Untested MilaX version found: ${MILAX_VERSION} -- assuming this version is compatible to MilaX version 0.4"
      MILAX_VERSION=0.4
    else
      echo "WARNING: Untested MilaX version found: ${MILAX_VERSION} -- assuming this version is compatible to MilaX version 0.3.3"
      MILAX_VERSION=0.3.3
    fi
    ;;

esac


# directory with the additional files
#
home=/usr/dss/share/hdinstall


# added 09.06.2009 22:11 /bs  -- start
umount /lib/libc.so.1
# added 09.06.2009 22:11 /bs  -- end

# start copying main data over to it
echo "Starting to copy data from UFS root to ${TARGET_FS} - this may take some time."
cd /
find . -xdev -depth -print | cpio -pvdm ${TARGET_FS} 2>&1 > /dev/null
# libc!
cp -rP /lib/libc.so* ${TARGET_FS}/lib 2>&1 > /dev/null

mkdir -p ${TARGET_FS}/alex
cd /alex
find . -xdev -depth -print | cpio -pvdm ${TARGET_FS}/alex 2>&1 > /dev/null
mkdir -p ${TARGET_FS}/root
cd /root
find . -xdev -depth -print | cpio -pvdm ${TARGET_FS}/root 2>&1 > /dev/null

mkdir -p ${TARGET_FS}/usr
cd /usr
find . -xdev -depth -print | cpio -pvdm ${TARGET_FS}/usr 2>&1 > /dev/null

mkdir -p ${TARGET_FS}/system/object
mkdir -p ${TARGET_FS}/system/contract
mkdir -p ${TARGET_FS}/tmp
mkdir -p ${TARGET_FS}/mnt
mkdir -p ${TARGET_FS}/proc
mkdir -p ${TARGET_FS}/var/run
mkdir -p ${TARGET_FS}/dev/fd
touch ${TARGET_FS}/etc/mnttab
touch ${TARGET_FS}/etc/dfs/sharetab

chmod 555 ${TARGET_FS}/system/object
chmod 555 ${TARGET_FS}/system/contract
chmod 555 ${TARGET_FS}/proc
chmod 777 ${TARGET_FS}/tmp
chmod 755 ${TARGET_FS}/etc/svc/volatile
chmod 755 ${TARGET_FS}/var/run
chmod 555 ${TARGET_FS}/dev/fd
chmod 444 ${TARGET_FS}/etc/mnttab
chmod 755 ${TARGET_FS}/usr
chmod 755 ${TARGET_FS}/alex
chmod 755 ${TARGET_FS}/root
chmod 755 ${TARGET_FS}/mnt
chown -R alex:adm ${TARGET_FS}/alex
chown root ${TARGET_FS}/system/object
chown root ${TARGET_FS}/system/contract
chown root ${TARGET_FS}/proc
chown root ${TARGET_FS}/tmp
chown root ${TARGET_FS}/etc/svc/volatile
chown root ${TARGET_FS}/var/run
chown root ${TARGET_FS}/dev/fd
chown root ${TARGET_FS}/etc/mnttab

chgrp root ${TARGET_FS}/system/object
chgrp root ${TARGET_FS}/system/contract
chgrp sys ${TARGET_FS}/tmp
chgrp root ${TARGET_FS}/proc
chgrp sys ${TARGET_FS}/etc/svc/volatile
chgrp sys ${TARGET_FS}/var/run
chgrp root ${TARGET_FS}/dev/fd
chgrp root ${TARGET_FS}/etc/mnttab

# now populate the devices and /dev directories in ${TARGET_FS}
# do this by lofs mounting / and pulling the files directly
mkdir -p /zfs-root-tmp.$$
mount -F lofs -o nosub / /zfs-root-tmp.$$
(cd /zfs-root-tmp.$$; tar cf - devices dev ) | (cd ${TARGET_FS}; tar xfp -) 2>&1 > /dev/null
cd /
umount /zfs-root-tmp.$$
rm -rf /zfs-root-tmp.$$

# copy last etc
cd /etc
tar cf - . | (cd ${TARGET_FS}/etc ; tar xfp -) 2>&1 > /dev/null

if [ "${MILAX_VERSION}"x = "0.3.3"x ] ; then
  cp $home/misc/xstart ${TARGET_FS}/usr/dss/sbin/
  cp $home/misc/live-fs-root ${TARGET_FS}/lib/svc/method/
  cp $home/misc/repository.db ${TARGET_FS}/etc/svc/
elif [ "${MILAX_VERSION}"x = "0.4"x ] ; then
  rm ${TARGET_FS}/lib/svc/method/live-fs-root
  cd ${TARGET_FS}/lib/svc/method/
  ln -s fs-root live-fs-root
  cd /
fi

# copy vfstab
cp $home/misc/vfstab ${TARGET_FS}/etc/vfstab


echo "/devices/ramdisk:a	-	/	ufs	1	no	rw,notime">>${TARGET_FS}/etc/vfstab


echo " ###################################################################"
echo
echo "       		MilaX now copied to ${TARGET_FS}"
echo
echo " ###################################################################"