See the notes in the beginning of each script for a summary of what it does.
This script, run at backup time, creates scripts similar to make.dev.hda and mount.dev.x, below, for you to run at restore time. It also produces data files similar to dev.hda, below. The name of the script and data file produced depends on the device given this script as a a parameter. That script, run at restore time, builds the partitions on the hard drive. make.fdisk is called from save.metadata, below.
#! /usr/bin/perl # A perl script to create a script and input file for fdisk to # re-create the partitions on the hard disk, and format the Linux and # Linux swap partitions. The first parameter is the fully qualified # path of the device of the hard disk, e.g. /dev/hda. The two # resulting files are the script make.dev.x and the data file dev.x # (where x is the hard drive described, e.g. hda, sdc). make.dev.x is # run at restore time to rebuild hard drive x, prior to running # restore.metadata. dev.x is the input file for fdisk. # Time-stamp: <2003-01-19 17:23:14 ccurley make.fdisk> # Copyright 2001 through the last date of modification Charles Curley # except for the subroutine cut2fmt. # cut2fmt Copyright (c) 1998 Tom Christiansen, Nathan Torkington and # O'Reilly & Associates, Inc. Permission is granted to use this code # freely EXCEPT for book publication. You may use this code for book # publication only with the explicit permission of O'Reilly & # Associates, Inc. # This program is free software; you can redistribute it and/or modify it # under the terms of the GNU General Public License as published by the # Free Software Foundation; either version 2 of the License, or (at your # option) any later version. # This program is distributed in the hope that it will be useful, but # WITHOUT ANY WARRANTY; without even the implied warranty of # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU # General Public License for more details. # You should have received a copy of the GNU General Public License along # with this program; if not, write to the Free Software Foundation, Inc., # 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA # In addition, as a special exception, Tom Christiansen, Nathan # Torkington and O'Reilly & Associates, Inc. give permission to use # the code of this program with the subroutine cut2fmt (or with # modified versions of the subroutine cut2fmt that use the same # license as the subroutine cut2fmt), and distribute linked # combinations including the two. You must obey the GNU General # Public License in all respects for all of the code used other than # the subroutine cut2fmt. If you modify this file, you may extend # this exception to your version of the file, but you are not # obligated to do so. If you do not wish to do so, delete this # exception statement and the subroutine cut2fmt from your version. # You can also contact the Free Software Foundation at http://www.fsf.org/ # Changes: # 2003 01 09: Added support for FAT32. We now create two scripts for # each hard drive, make.dev.[as]dx and mount.dev.[as]dx. These create # and make file systems on each partition, and make mount points and # mount them. # 2002 12 25: added support to handle W95 extended (LBA) (f) and W95 # FAT 32 partitions. I have tested this for primary but not logical # partitions. # 2002 09 08: Added minimal support for ext3fs. We now detect mounted # ext3fs partitions & rebuild but with no options. The detection # depends on the command line "dumpe2fs <device> 2>/dev/null | grep -i # journal" producing no output for an ext2fs, and output (we don't # care what) for an ext3fs. # This could stand extension to support non-default ext3 options such # as the type of journaling. Volunteers? # 2002 07 25: Bad block checking is now a command line option (-c) at # the time the product script is run. # 2002 07 03: Corrected the mechanism for specifying the default # drive. # 2001 11 25: Changed the way mke2fs gets its bad block # list. badblocks does not guess at the block size, so you have to get # it (from dumpe2fs) and feed it to badblocks. It is simpler to just # have mke2fs call badblocks, but you do loose the ability to have a # writing test easily. -- C^2 # 2001 11 25: Changed the regex that extracts partition labels from # the mount command. This change does not affect the results at all, # it just makes it possible to use Emacs' perl mode to indent # correctly. I just escaped the left bracket in the regex. -- C^2 # Discussion: # fdisk will spit out a file of the form below if you run it as "fdisk # -l". # root@tester ~/bin $ fdisk -l /dev/hda # Disk /dev/hda: 64 heads, 63 sectors, 1023 cylinders # Units = cylinders of 4032 * 512 bytes # Device Boot Start End Blocks Id System # /dev/hda1 1 9 18112+ 83 Linux # /dev/hda2 10 1023 2044224 5 Extended # /dev/hda5 10 368 723712+ 83 Linux # /dev/hda6 369 727 723712+ 83 Linux # /dev/hda7 728 858 264064+ 83 Linux # /dev/hda8 859 989 264064+ 83 Linux # /dev/hda9 990 1022 66496+ 82 Linux swap # What fdisk does not do is provide output suitable for later # importing into fdisk, a la sfdisk. This script parses the output # from fdisk and creates an input file for fdisk. Use the input file # like so: # fdisk /dev/hdx < dev.hdx # For the bare metal restore package, this script also builds a script # that will execute the above command so you can run it from your zip # disk. Because the bare metal restore scripts all are in /root/bin, # the data file and script created by this script are also placed # there. The same script also creates appropriate Linux file systems, # either ext2fs, or Linux swap. There is limited support for FAT12, # FAT16 and FAT32. For anything else, you're on your own. # Note for FAT32: According to the MS KB, there are more than one # reserved sectors for FAT32, usually 32, but it can vary. Do a search # in M$'s KB for "boot sector" or BPB for the gory details. For more # info than you really need on how boot sectors are used, see # http://support.microsoft.com/support/kb/articles/Q140/4/18.asp # You can also edit dev.x to change the sizes of partitions. Don't # forget, if you change the size of a FAT partition across the 32MB # boundary, you need to change the type as well! Run "fdisk /dev/hda" # or some such, then the l command to see the available partition # types. Then go ahead and edit dev.x appropriately. Also, when moving # partition boundarys with hand edits, make sure you move both logical # and extended partition boundaries appropriately. # Bad block checking right now is a quick read of the partition. A # writing check is also possible but more difficult. You have to run # badblocks as a separate command, and pass the bad block list to # mke2fs in a file (in /tmp, which is a ram disk). You also have to # know how large the blocks are, which you learn by running # dumpe2fs. It gets messy and I haven't done it yet. You probably # don't need it for a new hard drive, but if you have had a hard drive # crash on you and you are reusing it (while you are waiting for its # replacement to come in, I presume), then I highly recommend it. Let # me know how you do it. # For more information contact the author, Charles Curley, at # http://w3.trib.com/~ccurley/. # cut2fmt figures out the format string for the unpack function we use # to slice and dice the output from fdisk. From Christiansen and # Torkington, Perl Cookbook 5. sub cut2fmt { my (@positions) = @_; my $template = ''; my $lastpos = 1; foreach $place (@positions) { $template .= "A" . ($place - $lastpos) . " "; $lastpos = $place; } $template .= "A*"; return $template; } # Sub gpl, a subroutine to ship the GPL and other header information # to the current output file. sub gpl { print OUTPUT <<FINIS; # Copyright 2001 through the last date of modification Charles Curley. # This program is free software; you can redistribute it and/or modify it # under the terms of the GNU General Public License as published by the # Free Software Foundation; either version 2 of the License, or (at your # option) any later version. # This program is distributed in the hope that it will be useful, but # WITHOUT ANY WARRANTY; without even the implied warranty of # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU # General Public License for more details. # You should have received a copy of the GNU General Public License along # with this program; if not, write to the Free Software Foundation, Inc., # 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA # You can also contact the Free Software Foundation at http://www.fsf.org/ # For more information contact the author, Charles Curley, at # http://w3.trib.com/~ccurley/. FINIS } # Begin main line code. # Provide a default device. # print "\$ARGV[0] is $ARGV[0].\n"; $device = defined ($ARGV[0]) ? $ARGV[0] : "/dev/hda"; # print "Device is $device.\n"; # prepare format string. $fmt = cut2fmt (11, 19, 24, 34, 45, 49); # print "Format string is $fmt.\n\n"; # define fields in the array @_. $dev = 0; $bootable = 1; $firstcyl = 2; $lastcyl = 3; $parttype = 5; $partstring = 6; $target = "\/target"; $outputfilename = $device; $outputfilename =~ s/\//./g; $outputfilename = substr ($outputfilename, 1, 100); $outputfilepath = "\/root\/bin\/"; # Make a hash of the labels. $mpid = open (MOUNT, "mount -l |") or die "Couldn't fork: $!\n"; while (<MOUNT>) { if ($_ =~ /^$device/i) { # is this a line with a partition in it? # print $_; # print it just for grins split; if ($_[6] ne "") { # only process if there actually is a label $_[6] =~ s/[\[\]]//g; # strike [ and ]. $labels{$_[0]} = $_[6]; # print "The label of file device $_[0] is $labels{$_[0]}.\n"; } # We only mount if it's ext2fs or ext3fs and read and write. if ($_[4] =~ /ext[23]/ and $_[5] eq "(rw)" ) { $mountpoints{$_[2]} = $_[0]; # print "$_[2] is the mountpoint for $mountpoints{$_[2]}.\n"; } } } close (MOUNT); $fpid = open (FDISK, "fdisk -l $device |") or die "Couldn't fork: $!\n"; open (OUTPUT, "> $outputfilepath${outputfilename}") or die "Couldn't open output file $outputfilepath${outputfilename}.\n"; while (<FDISK>) { if ($_ =~ /^$device/i) { # is this a line with a partition in it? # print $_; # print it just for grins chop; # kill trailing \r @_ = unpack ($fmt, $_); # now strip white spaces from cylinder numbers @_[$firstcyl] =~ s/[ \t]+//; @_[$lastcyl] =~ s/[ \t]+//; @_[$parttype] =~ s/[ \t]+//; $partnumber = substr(@_[$dev], 8, 10); # get partition number for this line # just for grins # print " $partnumber, @_[$firstcyl], @_[$lastcyl], @_[$parttype], @_[$partstring]\n"; # Here we start creating the input to recreate the partition # this line represents. print OUTPUT "n\n"; if ($partnumber < 5) { # primary Linux partition if (@_[$parttype] == 83) { print OUTPUT "p\n$partnumber\n@_[$firstcyl]\n"; if (@_[$firstcyl] ne @_[$lastcyl]) { # in case it's all on one cylinder print OUTPUT "@_[$lastcyl]\n"; } # Now detect if this is an ext3 (journaling) # partition. We do this using dumpe2fs to dump the # partition and grepping on "journal". If the # partition is ext2, there will be no output. If it is # ext3, there will be output, and we use that fact to # set a command line switch. The command line switch # goes into an associative array (hash) so we don't # have to remember to reset it to the null string when # we're done. $dpid = open (DUMPE2FS, "dumpe2fs @_[$dev] 2>/dev/null | grep -i journal |") or die "Couldn't fork: $!\n"; while (<DUMPE2FS>) { # print "Dumpe2fs: $_"; $ext3{$_[$dev]} = "-j "; last; } close (DUMPE2FS); if ($labels{@_[$dev]}) { # do we have a label? $format .= "echo\necho formatting $checking@_[$dev]\n"; $format .= "mke2fs $ext3{$_[$dev]}\$blockcheck -L $labels{@_[$dev]} @_[$dev]\n\n"; } else { $format .= "echo\necho formatting $checking@_[$dev]\n"; $format .= "mke2fs $ext3{$_[$dev]}\$blockcheck @_[$dev]\n\n"; } # extended partition } elsif (@_[$parttype] == 5) { # print ("Creating Extended Partition.\n"); print OUTPUT "e\n$partnumber\n@_[$firstcyl]\n"; if (@_[$firstcyl] ne @_[$lastcyl]) { print OUTPUT "@_[$lastcyl]\n"; } # extended partition, Win95 Ext'd (LBA) } elsif (@_[$parttype] eq "f") { # print ("Creating Extended LBA Partition.\n"); print OUTPUT "e\n$partnumber\n@_[$firstcyl]\n"; if (@_[$firstcyl] ne @_[$lastcyl]) { print OUTPUT "@_[$lastcyl]\n"; } print OUTPUT "t\n$partnumber\nf\n"; # primary Linux swap partition } elsif (@_[$parttype] == 82) { print OUTPUT "p\n$partnumber\n@_[$firstcyl]\n"; if (@_[$firstcyl] ne @_[$lastcyl]) { print OUTPUT "@_[$lastcyl]\n"; } print OUTPUT "t\n$partnumber\n82\n"; $format .= "echo Making @_[$dev] a swap partition.\n"; $format .= "mkswap \$blockcheck @_[$dev]\n\n"; # Primary mess-dos partition. We don't handle hidden # partitions. } elsif ( @_[$parttype] == 1 || @_[$parttype] == 4 || @_[$parttype] == 6 || @_[$parttype] eq "b" || @_[$parttype] eq "c" || @_[$parttype] eq "e" ) { # print ("Making DOS primary partition.\n"); print ("dd if=@_[$dev] of=$outputfilepath$outputfilename$partnumber"); print (" bs=512 count=1\n"); system ("dd if=@_[$dev] of=$outputfilepath$outputfilename$partnumber bs=512 count=1"); print OUTPUT "p\n$partnumber\n@_[$firstcyl]\n"; if (@_[$firstcyl] ne @_[$lastcyl]) { # in case it's all on one cylinder print OUTPUT "@_[$lastcyl]\n"; } print OUTPUT "t\n$partnumber\n@_[$parttype]\n"; $format .= "echo\necho formatting $checking@_[$dev]\n"; $format .= "mkdosfs \$blockcheck"; if ( @_[$parttype] == b || @_[$parttype] == c) { # We have a W9x FAT32 partition. Add a command line switch. $format .= " -F 32"; } $format .= " @_[$dev]\n"; $format .= "# restore FAT boot sector.\n"; $format .= "dd if=$outputfilename$partnumber of=@_[$dev] bs=512 count=1\n\n"; } else { # anything else partition print OUTPUT "p\n@_[$firstcyl]\n"; if (@_[$firstcyl] ne @_[$lastcyl]) { print OUTPUT "@_[$lastcyl]\n"; } print OUTPUT "t\n$partnumber\n@_[$parttype]\n"; } } else { # logical Linux partition if (@_[$parttype] == 83) { print OUTPUT "l\n@_[$firstcyl]\n"; if (@_[$firstcyl] ne @_[$lastcyl]) { print OUTPUT "@_[$lastcyl]\n"; } # Now detect if this is an ext3 (journaling) # partition. We do this using dumpe2fs to dump the # partition and grepping on "journal". If the # partition is ext2, there will be no output. If it is # ext3, there will be output, and we use that fact to # set a command line switch. The command line switch # goes into an associative array (hash) so we don't # have to remember to reset it to the null string when # we're done. $dpid = open (DUMPE2FS, "dumpe2fs @_[$dev] 2>/dev/null | grep -i journal |") or die "Couldn't fork: $!\n"; while (<DUMPE2FS>) { # print "Dumpe2fs: $_"; $ext3{$_[$dev]} = "-j "; last; } close (DUMPE2FS); if ($labels{@_[$dev]}) { # do we have a label? $format .= "echo\necho formatting $checking@_[$dev]\n"; $format .= "mke2fs $ext3{@_[$dev]}\$blockcheck -L $labels{@_[$dev]} @_[$dev]\n\n"; } else { $format .= "echo\necho formatting $checking@_[$dev]\n"; $format .= "mke2fs $ext3{@_[$dev]}\$blockcheck @_[$dev]\n\n"; } # logical Linux swap partition } elsif (@_[$parttype] == 82 ) { print OUTPUT "l\n@_[$firstcyl]\n"; if (@_[$firstcyl] ne @_[$lastcyl]) { print OUTPUT "@_[$lastcyl]\n"; } print OUTPUT "t\n$partnumber\n82\n"; $format .= "echo Making @_[$dev] a swap partition.\n"; $format .= "mkswap \$blockcheck @_[$dev]\n\n"; # Logical mess-dos partition. We don't handle hidden # partitions. } elsif ( @_[$parttype] == 1 || @_[$parttype] == 4 || @_[$parttype] == 6 || @_[$parttype] eq "b" || @_[$parttype] eq "c" || @_[$parttype] eq "e" ) { # print ("Making DOS logical partition.\n"); print ("dd if=@_[$dev] of=$outputfilepath$outputfilename$partnumber"); print (" bs=512 count=1\n"); system ("dd if=@_[$dev] of=$outputfilepath$outputfilename$partnumber bs=512 count=1"); print OUTPUT "l\n$partnumber\n@_[$firstcyl]\n"; if (@_[$firstcyl] ne @_[$lastcyl]) { # in case it's all on one cylinder print OUTPUT "@_[$lastcyl]\n"; } print OUTPUT "t\n$partnumber\n@_[$parttype]\n"; $format .= "echo\necho formatting $checking@_[$dev]\n"; $format .= "mkdosfs \$blockcheck"; if ( @_[$parttype] == b || @_[$parttype] == c) { # We have a W9x FAT32 partition. Add a command line switch. $format .= " -F 32"; } $format .= " @_[$dev]\n"; $format .= "# restore FAT boot sector.\n"; $format .= "dd if=$outputfilename$partnumber of=@_[$dev] bs=512 count=1\n\n"; } else { # anything else partition print OUTPUT "l\n@_[$firstcyl]\n"; if (@_[$firstcyl] ne @_[$lastcyl]) { print OUTPUT "@_[$lastcyl]\n"; } print OUTPUT "t\n$partnumber\n@_[$parttype]\n"; } } # handle bootable partitions if (@_[$bootable] =~ /\*/) { print OUTPUT "a\n$partnumber\n"; } } } print OUTPUT "v\nw\n"; close (OUTPUT); close (FDISK); open (OUTPUT, "> ${outputfilepath}make.$outputfilename") or die "Couldn't open output file ${outputfilepath}make.$outputfilename.\n"; print OUTPUT <<FINIS; #! /bin/sh # A script to restore the partition data of a hard drive and format # the partitions. Created at bare metal backup time by the Perl script # make.fdisk. FINIS &gpl; print OUTPUT <<FINIS; export blockcheck=\$1; if [ "\$blockcheck" != "-c" ] && [ -n "\$blockcheck" ] then echo "\${0}: automated restore with no human interaction." echo "\${0}: -c: block check during file system making." exit 1; fi FINIS # Clean the old partition table out. print OUTPUT "dd if=/dev/zero of=$device bs=512 count=2\n\nsync\n\n"; print OUTPUT "fdisk $device \< $outputfilename\n\nsync\n\n"; print OUTPUT $format; print OUTPUT "fdisk -l \"$device\"\n"; close (OUTPUT); # Now build the script that will build the mount points on the root # and other partitions. open (OUTPUT, "> ${outputfilepath}mount.$outputfilename") or die "Couldn't open output file ${outputfilepath}make.$outputfilename.\n"; print OUTPUT <<FINIS; #! /bin/sh # A script to create a minimal directory tree on the target hard drive # and mount the partitions on it. Created at bare metal backup time by # the Perl script make.fdisk. FINIS &gpl; print OUTPUT <<FINIS; # WARNING: If your Linux system mount partitions across hard drive # boundaries, you will have multiple "mount.dev.* scripts. You must # ensure that they run in the proper order. The root partition should # be mounted first, then the rest in the order they cascade. If they # cross mount, you'll have to handle that manually. FINIS # We have a hash of mount points and devices in %mountpoints. However, # we have to process them such that directories are built on the # appropriate target partition. E.g. where /usr/local is on its own # partition, we have to mount /usr before we build /usr/local. We can # ensure this by sorting them. Shorter mount point paths will be built # first. We can't sort a hash directly, so we use an array. # We build commands to create the appropriate mount points and then # mount the partitions to the mount points. This is in preparation for # untarring the contents of the ZIP disk, done in restore.metadata. foreach $point ( sort keys %mountpoints) { print OUTPUT "\n# $point is the mountpoint for $mountpoints{$point}.\n"; print OUTPUT "mkdir $target$point\n"; print OUTPUT "mount $mountpoints{$point} $target$point\n"; } print OUTPUT "\nmount | grep -i \"$device\"\n"; close (OUTPUT); # These scripts are dangerous & should only be visible to root. chmod 0700, "${outputfilepath}make.$outputfilename"; chmod 0700, "${outputfilepath}mount.$outputfilename"; chmod 0600, "${outputfilepath}$outputfilename"; |
This script is a sample of the sort produced by make.fdisk, above. It uses data files like dev.hda, below. It builds partitions and puts file systems on some of them. This is the first script run at restore time.
If you are brave enough to edit dev.hda (q.v.), say, to add a new partition, you may need to edit this script as well.
If you want make.dev.hda to check for bad blocks when it puts a file system on the partitions, use a "-c" command line option.
#! /bin/sh # A script to restore the partition data of a hard drive and format # the partitions. Created at bare metal backup time by the Perl script # make.fdisk. # Copyright 2001 through the last date of modification Charles Curley. # This program is free software; you can redistribute it and/or modify it # under the terms of the GNU General Public License as published by the # Free Software Foundation; either version 2 of the License, or (at your # option) any later version. # This program is distributed in the hope that it will be useful, but # WITHOUT ANY WARRANTY; without even the implied warranty of # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU # General Public License for more details. # You should have received a copy of the GNU General Public License along # with this program; if not, write to the Free Software Foundation, Inc., # 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA # You can also contact the Free Software Foundation at http://www.fsf.org/ # For more information contact the author, Charles Curley, at # http://w3.trib.com/~ccurley/. export blockcheck=$1; if [ "$blockcheck" != "-c" ] && [ -n "$blockcheck" ] then echo "${0}: automated restore with no human interaction." echo "${0}: -c: block check during file system making." exit 1; fi dd if=/dev/zero of=/dev/hda bs=512 count=2 sync fdisk /dev/hda < dev.hda sync echo echo formatting /dev/hda1 mkdosfs $blockcheck /dev/hda1 # restore FAT boot sector. dd if=dev.hda1 of=/dev/hda1 bs=512 count=1 echo echo formatting /dev/hda2 mke2fs -j $blockcheck -L /boot /dev/hda2 echo echo formatting /dev/hda3 mke2fs -j $blockcheck -L / /dev/hda3 echo Making /dev/hda5 a swap partition. mkswap $blockcheck /dev/hda5 fdisk -l "/dev/hda" |
This script is a sample of the sort produced by make.fdisk, above. It builds mount points and mounts partitions on them, making the target file system ready for restoring files. This is the second script run at restore time.
If you are brave enough to edit dev.hda (q.v.), say, to add a new partition, you may need to edit this script as well.
#! /bin/sh # A script to create a minimal directory tree on the target hard drive # and mount the partitions on it. Created at bare metal backup time by # the Perl script make.fdisk. # Copyright 2001 through the last date of modification Charles Curley. # This program is free software; you can redistribute it and/or modify it # under the terms of the GNU General Public License as published by the # Free Software Foundation; either version 2 of the License, or (at your # option) any later version. # This program is distributed in the hope that it will be useful, but # WITHOUT ANY WARRANTY; without even the implied warranty of # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU # General Public License for more details. # You should have received a copy of the GNU General Public License along # with this program; if not, write to the Free Software Foundation, Inc., # 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA # You can also contact the Free Software Foundation at http://www.fsf.org/ # For more information contact the author, Charles Curley, at # http://w3.trib.com/~ccurley/. # WARNING: If your Linux system mount partitions across hard drive # boundaries, you will have multiple "mount.dev.* scripts. You must # ensure that they run in the proper order. The root partition should # be mounted first, then the rest in the order they cascade. If they # cross mount, you'll have to handle that manually. # / is the mountpoint for /dev/hda3. mkdir /target/ mount /dev/hda3 /target/ # /boot is the mountpoint for /dev/hda2. mkdir /target/boot mount /dev/hda2 /target/boot mount | grep -i "/dev/hda" |
This data file is used at restore time. It is fed to fdisk by the script make.dev.hda. It is produced at backup time by make.fdisk. Those familiar with fdisk will recognize that each line is an fdisk command or value, such as a cylinder number. Thus, it is possible to change the partition sizes and add new partitions by editing this file. That's why the penultimate command is v, to verify the partition table before it is written.
n p 1 1 29 t 1 6 a 1 n p 2 30 44 n e 3 45 1023 n l 45 944 n l 945 1023 t 6 82 v w |
This is the first script to run as part of the backup process. It calls make.fdisk, above. If you have a SCSI hard drive or multiple hard drives to back up, edit the call to make.fdisk appropriately.
#! /bin/sh # A script to save certain meta-data off to the boot partition. Useful for # restoration. # Time-stamp: <2003-01-19 18:12:29 root save.metadata> # Copyright 2000 through the last date of modification, Charles Curley. # This program is free software; you can redistribute it and/or modify it # under the terms of the GNU General Public License as published by the # Free Software Foundation; either version 2 of the License, or (at your # option) any later version. # This program is distributed in the hope that it will be useful, but # WITHOUT ANY WARRANTY; without even the implied warranty of # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU # General Public License for more details. # You should have received a copy of the GNU General Public License along # with this program; if not, write to the Free Software Foundation, Inc., # 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA # You can also contact the Free Software Foundation at http://www.fsf.org/ # 2003 01 08: We now age the output from rpm -VA to make back # comparisons easier. # The loop that creates directories now has the -p option for mkdir, # which means you can create parents on the fly if they don't already # exist. # initrd is now in the list of directories to create automatically. # We now exclude more stuff when building the tarballs. # 2002 07 01: Went to bzip2 to compress the archives, for smaller # results. This is important in a 100MB ZIP disk. Also some general # code cleanup. # 2002 07 01: The function crunch will tar and BZIP2 the # archives. This is cleaner than the old code, and has better safety # checking. # For more information contact the author, Charles Curley, at # http://w3.trib.com/~ccurley/. # Crunch: A function to compress the contents of a directory and put # the archive onto the ZIP disk. # The first parameter is the name of the archive file to be # created. The backup location, $zip, will be prepended and the # extension, "tar.bz2" will be appended. # All following parameters will be taken as additional directories or # files to be put into the archive. function crunch { if [ -z "$1" ] || [ -z "$2" ] # Checks if parameter #1 or #2 is zero length. then echo "-Parameter #1 or #2 is missing.-" # Also if no parameter is passed. return 1 else local file=$1 # The archive file to create shift # Discard the file name local dirs=$@ # The director[y|ies] to archive local tarcmd="tar -cjf" # The tar command. $tarcmd $zip/$file.tar.bz2 $dirs # do it!! error=$? # Preserve the exit code if [ $error != 0 ] # Did we fail? then # Yes echo "Tar failed with error $error" echo $tarcmd $zip/$file.tar.bz2 $dirs exit $error # return tar's exit code as ours fi return 0 # For error testing if needed. fi } # Begin the main line code export zip="/mnt/zip"; # Where we will put archives # export save="/mnt/save"; RPMVABACKS=/etc # where we keep our backups RPMVAROOT=rpmVa # The root name of the pg backups ANC=${RPMVABACKS}/${RPMVAROOT}.anc # name for the oldest (ancient) backup OLD=${RPMVABACKS}/${RPMVAROOT}.old # name for the middling oldest backup NEW=${RPMVABACKS}/${RPMVAROOT}.txt # name for the newest backup if [ -f ${ANC} ]; then echo "Deleting ${ANC}" rm ${ANC} fi if [ -f ${OLD} ]; then echo "Aging ${OLD}" mv ${OLD} ${ANC} fi if [ -f ${NEW} ]; then echo "Aging ${NEW}" mv ${NEW} ${OLD} fi # Now we save hard drive information. Run make.fdisk on each hard # drive in the order in which it mounted from the root partition. That # is, run it first on the hard drive with your root partition, then # any hard drives that mount to the first hard drive, then any hard # drives that mount to those. For example, if your root partition is # on /dev/sdc, run "make.fdisk /dev/sdc" first. # The reason for this is that make.fdisk produces a script to make # mount points and then mount the appropriate partition to them during # first stage restore. Mount points must be created on the partition # where they will reside. The partitions must be mounted in this # order. For example, if your /var and /var/ftp are both separate # partitions, then you must mount /, create /var, then mount /var, # then create /var/ftp. The order in which the script "first.stage" # runs the mounting scripts is based on their time of creation. # If necessary, put a line, "sleep 1" between calls to make.fdisk. echo "Saving hard drive info" make.fdisk /dev/hda # back up RPM metadata echo "Verifying RPMs." rpm -Va | sort +2 -t ' ' > ${NEW} echo "Finished verifying RPMs; now mounting the ZIP drive." # Make sure we have the ZIP drive mounted. umount $zip modprobe ppa # Driver for 100MB parallel port ZIP disk mount $zip # It should have ext2fs on partition 1. # clean it all out rm -r $zip/* mkdir $zip/lost+found echo "`hostname` bare metal ZIP disk, created `date`" > $zip/README.txt echo "Building the ZIP drive backups." # These are in case we need to refer to them while rebuilding. The # rebuilding process should be mostly automated, but you never # know.... fdisk -l /dev/hda > $zip/fdisk.hda ls -al /mnt > $zip/ls.mnt.txt ls -al / > $zip/ls.root.txt cd / # Build our minimal archives on the ZIP disk. These appear to be # required so we can restore later on. crunch boot boot crunch root root --exclude boot/.cpan --exclude root/.mozilla crunch etc etc --exclude etc/samba --exclude X11 crunch lib lib crunch usr.sbin usr/sbin crunch usr.bin usr/bin --exclude usr/bin/emacs --exclude usr/bin/emacs-21.2 --exclude usr/bin/emacsclient --exclude usr/bin/emacs-nox --exclude usr/bin/gs --exclude usr/bin/pine --exclude usr/bin/gimp-1.2 --exclude usr/bin/doxygen --exclude usr/bin/postgres --exclude usr/bin/gdb --exclude usr/bin/kmail --exclude usr/bin/splint --exclude usr/bin/odbctest --exclude usr/bin/php --exclude usr/bin/xchat --exclude usr/bin/gnucash --exclude usr/bin/pdfetex --exclude usr/bin/pdftex --exclude usr/bin/smbcacls --exclude usr/bin/evolution-calendar --exclude usr/bin/xpdf --exclude usr/bin/xmms crunch sbin sbin crunch bin bin crunch dev dev crunch kerberos usr/kerberos/lib/ # Now optional saves. # arkeia specific: # crunch arkeia usr/knox # save these so we can use ssh for restore. *crack* for RH 7.0 login # authentication. crunch usr.lib usr/lib/*crack* usr/lib/libz* usr/lib/libssl* usr/lib/libcrypto* # save the scripts we used to create the ZIP disk and the ones we will # use to restore it. mkdir $zip/root.bin cp -p /root/bin/* $zip/root.bin rm $zip/root.bin/*~ $zip/root.bin/#*# echo "Testing our results." find $zip -iname "*.bz2" | xargs bunzip2 -t # Not a normal part of the process: we duplicate the ZIP disk onto an # NFS mount elsewhere. # echo "Backing the ZIP drive to the NFS mount." # umount $save # mount $save # rm -r $save/zip # mkdir $save/zip # cp -pr $zip $save du -hs ${zip}* df -m |
This script restores metadata from the ZIP disk as a first stage restore.
#! /bin/sh # A script to restore the meta-data from the ZIP disk. This runs under # tomsrtbt only after partitions have been rebuilt, file systems made, # and mounted. It also assumes the ZIP disk has already been # mounted. Mounting the ZIP disk read only is probably a good idea. # Time-stamp: <2003-01-08 18:26:44 ccurley restore.metadata> # Copyright 2000 through the last date of modification Charles Curley. # This program is free software; you can redistribute it and/or modify it # under the terms of the GNU General Public License as published by the # Free Software Foundation; either version 2 of the License, or (at your # option) any later version. # This program is distributed in the hope that it will be useful, but # WITHOUT ANY WARRANTY; without even the implied warranty of # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU # General Public License for more details. # You should have received a copy of the GNU General Public License along # with this program; if not, write to the Free Software Foundation, Inc., # 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA # You can also contact the Free Software Foundation at http://www.fsf.org/ # 2002 07 01: Went to bzip2 to compress the archives, for smaller # results. This is important in a 100MB ZIP disk. Also some general # code cleanup. # For more information contact the author, Charles Curley, at # http://w3.trib.com/~ccurley/. zip="/mnt"; # Where we mount the zip drive. target="/target"; # Where the hard drive to restore is mounted. ls -lt $zip # Warm fuzzies for the user. cd $target # Restore the archived metadata files. for archive in $( ls $zip/*.bz2 ); do echo $archive ls -al $archive bzip2 -dc $archive | tar -xf - done # Build the mount points for our second stage restoration and other # things. # If you boot via an initrd, make sure you build a directory here so # the kernel can mount the initrd at boot. for dir in mnt/save mnt/zip mnt/cdrom mnt/floppy mnt/imports proc initrd; do mkdir -p $target/$dir done chmod a-w $target/proc # Restore /proc's read-only permissions # Restore the scripts we used to create the ZIP disk and the ones we will # use to restore it. These should be the latest & greatest in case we had # to do any editing during 1st stage restore. cp -p $zip/root.bin/* $target/root/bin # Now install the boot sector. chroot $target /sbin/lilo -C /etc/lilo.conf df -m |
This script runs the entire first stage restore with no operator intervention.
If you want to check for bad blocks when it puts a file system on the partitions, use a "-c" command line option.
#! /bin/sh # A master script to run the other, detailed scripts. Use this script # only if you want no human intervention in the restore process. The # only option is -c, which forces bad block checking during formatting # of the partitions. # Time-stamp: <2003-01-09 11:32:59 ccurley first.stage> # Copyright 2002 through the last date of modification Charles Curley. # This program is free software; you can redistribute it and/or modify it # under the terms of the GNU General Public License as published by the # Free Software Foundation; either version 2 of the License, or (at your # option) any later version. # This program is distributed in the hope that it will be useful, but # WITHOUT ANY WARRANTY; without even the implied warranty of # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU # General Public License for more details. # You should have received a copy of the GNU General Public License along # with this program; if not, write to the Free Software Foundation, Inc., # 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA # You can also contact the Free Software Foundation at http://www.fsf.org/ export blockcheck=$1; if [ "$blockcheck" != "-c" ] && [ -n "$blockcheck" ] then echo "${0}: automated restore with no human interaction." echo "${0}: -c: block check during file system making." exit 1; fi cd /mnt/root.bin # just in case we aren't already where we should be. for drive in $( ls make.dev.* ); do ./$drive $blockcheck; done # WARNING: If your Linux system mount partitions across hard drive # boundaries, you will have multiple "mount.dev.* scripts. You must # ensure that they run in the proper order, which the loop below may # not do. The root partition should be mounted first, then the rest in # the order they cascade. If they cross mount, you'll have to handle # that manually. # The "ls -tr" will list the scripts in the order they are created, so # it might be a good idea to create them (in the script save.metadata) # in the order in which you should run them. for drive in $( ls -tr mount.dev.* ); do ./$drive; done ./restore.metadata # People who are really confident may comment this line in. # reboot |
These scripts run on the computer being backed up or restored.
This script saves to another computer via an NFS mount. You can adapt it to save to tape drives or other media.
#! /bin/sh # Back up the entire system to another computer's drive. To make this # work, we need a convenient chunk of disk space on the remote computer we # can nfs mount as /mnt/save. # Time-stamp: <2001-11-19 09:19:35 ccurley back.up.all> # Copyright 2000 through the last date of modification Charles Curley. # This program is free software; you can redistribute it and/or modify it # under the terms of the GNU General Public License as published by the # Free Software Foundation; either version 2 of the License, or (at your # option) any later version. # This program is distributed in the hope that it will be useful, but # WITHOUT ANY WARRANTY; without even the implied warranty of # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU # General Public License for more details. # You should have received a copy of the GNU General Public License along # with this program; if not, write to the Free Software Foundation, Inc., # 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA # You can also contact the Free Software Foundation at http://www.fsf.org/ # For more information contact the author, Charles Curley, at # http://w3.trib.com/~ccurley/. save="/mnt/save" # Make sure it's there umount $save mount $save cd / rm $save/tester.tar.old.gz mv $save/tester.tar.gz $save/tester.tar.old.gz # save everything except /mnt, /proc, and nfs mounted directories. time tar cf - / --exclude /mnt --exclude /proc --exclude $save\ | gzip -c > $save/tester.tar.gz |
This script does exactly what back.up.all does, but it uses ssh instead of nfs.
#! /bin/sh # Back up the entire system to another computer's drive. To make this # work, we need a convenient chunk of disk space on the remote # computer. This version uses ssh to do its transfer, and compresses # using bz2. This means this script has to know more about the other # computer, which does not make for good modularization. # Time-stamp: <2001-11-19 09:19:48 ccurley back.up.all.ssh> # Copyright 2000 through the last date of modification Charles Curley. # This program is free software; you can redistribute it and/or modify it # under the terms of the GNU General Public License as published by the # Free Software Foundation; either version 2 of the License, or (at your # option) any later version. # This program is distributed in the hope that it will be useful, but # WITHOUT ANY WARRANTY; without even the implied warranty of # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU # General Public License for more details. # You should have received a copy of the GNU General Public License along # with this program; if not, write to the Free Software Foundation, Inc., # 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA # You can also contact the Free Software Foundation at http://www.fsf.org/ # For more information contact the author, Charles Curley, at # http://w3.trib.com/~ccurley/. save="/backs/tester" backup_server="charlesc" # rotate the old backups. Do it all in one line to minimze authentication overhead. ssh $backup_server "rm $save/tester.tar.old.bz2; mv $save/tester.tar.bz2 \ $save/tester.tar.old.bz2" # save everything except /mnt, /proc, and squid directories. time tar cf - / --exclude /mnt --exclude /proc --exclude /var/spool/squid\ | ssh $backup_server "bzip2 -9 > $save/tester.tar.bz2" |
This is the restore script to use if you backed up using back.up.all.
#! /bin/sh # A script to restore all of the data from an nfs mount. This is our final # stage restore. # Time-stamp: <2001-11-19 09:21:22 ccurley restore.all> # Copyright 2000 through the last date of modification Charles Curley. # This program is free software; you can redistribute it and/or modify it # under the terms of the GNU General Public License as published by the # Free Software Foundation; either version 2 of the License, or (at your # option) any later version. # This program is distributed in the hope that it will be useful, but # WITHOUT ANY WARRANTY; without even the implied warranty of # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU # General Public License for more details. # You should have received a copy of the GNU General Public License along # with this program; if not, write to the Free Software Foundation, Inc., # 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA # You can also contact the Free Software Foundation at http://www.fsf.org/ # For more information contact the author, Charles Curley, at # http://w3.trib.com/~ccurley/. export save="/mnt/save" mount $save cd / gunzip -dc $save/tester.tar.gz | tar -xpkf - rm /var/run/*.pid lilo |
This is the restoration script to use if you used back.up.all.ssh to back up.
#! /bin/sh # A script to restore all of the data using ssh and bunzip2. This is # our final stage restore. # Copyright 2000 through the last date of modification Charles Curley. # Time-stamp: <2001-11-19 09:21:33 ccurley restore.all.ssh> # This program is free software; you can redistribute it and/or modify it # under the terms of the GNU General Public License as published by the # Free Software Foundation; either version 2 of the License, or (at your # option) any later version. # This program is distributed in the hope that it will be useful, but # WITHOUT ANY WARRANTY; without even the implied warranty of # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU # General Public License for more details. # You should have received a copy of the GNU General Public License along # with this program; if not, write to the Free Software Foundation, Inc., # 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA # You can also contact the Free Software Foundation at http://www.fsf.org/ # For more information contact the author, Charles Curley, at # http://w3.trib.com/~ccurley/. save="/backs/tester/" backup_server="charlesc" cd / ssh $backup_server "cat $save/tester.tar.bz2" | bunzip2 | tar -xpkf - rm /var/run/*.pid lilo |
The ssh scripts above have a possible security problem. If you run them on a firewall, the firewall has to have access via ssh to the backup server. In that case, a clever cracker might also be able to crack the backup server. It would be more secure to run backup and restore scripts on the backup server, and let the backup server have access to the firewall. That is what these scripts are for. Rename them to get.x and restore.x where x is the name of the target computer. Edit them (the variable $target's initialization) to use the target computer's host name, or rewrite them to use a command line argument.
These scripts backup and restore the target completely, not just the stage one backup and restore. Also, note that get.tester backs up the ZIP disk as well, in case you need to replace a faulty ZIP disk.
I use these scripts routinely.
#! /bin/sh # Back up another computer's drive to this system. To make this work, we # need a convenient chunk of disk space on this computer. This version # uses ssh to do its transfer, and compresses using bz2. This version was # developed so that the system to be backed up won't be authenticated to # log onto the backup computer. This script is intended to be used on a # firewall. You don't want the firewall to be authenticated to the backup # system in case the firewall is cracked. # Time-stamp: <2002-08-23 16:12:35 ccurley get.tester> # Copyright 2000 through the last date of modification Charles Curley. # This program is free software; you can redistribute it and/or modify it # under the terms of the GNU General Public License as published by the # Free Software Foundation; either version 2 of the License, or (at your # option) any later version. # This program is distributed in the hope that it will be useful, but # WITHOUT ANY WARRANTY; without even the implied warranty of # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU # General Public License for more details. # You should have received a copy of the GNU General Public License along # with this program; if not, write to the Free Software Foundation, Inc., # 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA # You can also contact the Free Software Foundation at http://www.fsf.org/ # For more information contact the author, Charles Curley, at # http://w3.trib.com/~ccurley/. # 2002 07 01: We now set the path on the target to the zip drive with # a variable. This fixes a bug in the command to eject the zip disk. # 2002 07 01: The zip disk archives are now in bzip2 format, so this # script has been changed to reflect that. # The host name of the computer to be backed up. target=tester zip=/mnt/zip echo Backing up $target echo Aging the ZIP disk backups. rm -r $target.old.zip mv $target.zip $target.old.zip ssh $target "modprobe ppa ; mount -r $zip" echo Copying the ZIP disk. # -r for recursive copy, -p to preserve times and permissions, -q for # quiet: no progress meter. scp -qpr $target:$zip $target.zip du -hs $target.zip echo Aging the archives rm $target.tar.old.bz2 mv $target.tar.bz2 $target.tar.old.bz2 echo Backing up $target to the backup server. ssh $target tar -cf - / --exclude /mnt --exclude /proc --exclude /var/spool/squid\ | bzip2 -9 | cat > $target.tar.bz2 echo Testing the results. find . -iname "*.bz2" | xargs bunzip2 -t ssh $target "eject $zip" |
#! /bin/sh # A script to restore all of the data to tester via ssh. This is our final # stage restore. # Time-stamp: <2001-11-19 09:22:01 ccurley restore.tester> # Copyright 2000 through the last date of modification Charles Curley. # This program is free software; you can redistribute it and/or modify it # under the terms of the GNU General Public License as published by the # Free Software Foundation; either version 2 of the License, or (at your # option) any later version. # This program is distributed in the hope that it will be useful, but # WITHOUT ANY WARRANTY; without even the implied warranty of # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU # General Public License for more details. # You should have received a copy of the GNU General Public License along # with this program; if not, write to the Free Software Foundation, Inc., # 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA # You can also contact the Free Software Foundation at http://www.fsf.org/ # For more information contact the author, Charles Curley, at # http://w3.trib.com/~ccurley/. # The host name of the computer to be restored. target=tester bunzip2 -dc $target.tar.bz2 | ssh $target "cd / ; tar -xpkf - " ssh $target lilo |
Закладки на сайте Проследить за страницей |
Created 1996-2024 by Maxim Chirkov Добавить, Поддержать, Вебмастеру |