hcoop-backup: Take care of some files that don't need to be backed up

[clinton/scripts.git] / hcoop-backup

#!/bin/bash -e

#
# run this script as root, on deleuze, in a directory that WILL NOT be
# part of the backup (/tmp is a good choice).
#

# FIXME:
# postgres dumps
# mysql dumps

# for catsync
PATH=$PATH:/home/megacz_admin/
COMPRESS_EXT=.bz2
COMPRESS_PROG=bzip2

#KEYFILE=$1
KEYFILE=/etc/backup-encryption-key
BACKUPDIR=/afs/megacz.com/hcoop-backup/

cd $BACKUPDIR
find * -prune -ctime +5 -exec rm -rf {} \;

SUBDIR=`date +%Y.%m.%d`
rm -rf $SUBDIR
mkdir -p $SUBDIR
cd $SUBDIR

echo 'I am in:'
pwd
echo

echo building package lists...
dpkg-query -W -f='${Package}\n' > packages
(cd /; sudo find / /usr/ /usr/local/ /var/ -xdev)   | sort | uniq > allfiles
dpkg-query -W -f='${Package}\n'      | xargs dpkg -L | sort | uniq > debfiles
dpkg-query -W -f='${Conffiles}\n'           | grep / | cut -b2- | sed 's_ .*__' | sort | uniq > conffiles

diff allfiles debfiles | grep '^<' | cut -b 3- | \
 grep -v ^/var/cache | \
 grep -v ^/var/tmp | \
 grep -v ^/var/lib/dpkg | \
 grep -v ^/var/backups | \
 grep -v ^/var/lib/changetrack | \
 grep -v ^/var/run | \
 grep -v ^/var/lock | \
 grep -v ^/var/lib/ucf | \
 grep -v ^/vicepa | \
 grep -v ^/home | \
 grep -v ^/tmp | \
 grep -v '^/afs$' | \
 grep -v '^/$' | \
 grep -v '^/usr/$' | \
 grep -v ^/usr/src | \
 grep -v '^/usr/.*pyc' | \
 grep -v '^/usr/.*elc' | \
 grep -v '^/usr/lib/gconv/gconv-modules\.cache$' | \
 grep -v '^/usr/lib/locale/locale-archive$' | \
 grep -v '^/usr/share/info/dir$' | \
 grep -v '^/usr/share/info/dir\.old$' | \
 grep -v '^/usr/share/emacs21/site-lisp/' | \
 grep -v '^/usr/share/emacs22/site-lisp/' | \
 grep -v '^/usr/share/vim/addons/doc/tags$' | \
 cat > backupfiles

cat conffiles >> backupfiles

cat backupfiles | \
 grep -v ^/home | \
 grep -v ^/usr/local | \
 grep -v ^/var/spool | \
 grep -v ^/var/log | \
 grep -v ^/usr/lib/python2.4/ | \
 grep -v ^/var/lib/python-support | \
 grep -v ^/usr/share/man | \
 grep -v ^/media | \
 grep -v ^/vmlinuz | \
 grep -v ^/vmlinuz.old | \
 grep -v '^/sbin/[a-z\-]*.modutils$' | \
 cat > annoyingfiles-

for A in `cat annoyingfiles-`
do \
  test -L "$A" || echo "$A"
done > annoyingfiles

cat annoyingfiles | \
 grep -v ^/opt/dell/srvadmin/ | \
 grep -v ^/boot/ | \
 grep -v ^/dev/ | \
 grep -v ^/etc/ | \
 grep -v ^/root/ | \
 grep -v ^/var/ | \
 grep -v ^/lib/modules/ | \
 grep -v ^/var/domtool/ | \
 grep -v ^/var/lib/mysql/ | \
 grep -v ^/var/lib/postgres/ | \
 grep -v ^/var/lib/postgresql/ | \
 cat > complain

grep '[a-z/]' complain && \
  mail -a 'From: Adam Megacz <megacz@hcoop.net>' \
       -s "automated message: annoying files found on deleuze (please do something about them)" admins@hcoop.net \
       < complain

cp backupfiles /tmp/backupfiles
(cd /; sudo tar cvlpjf - --ignore-failed-read --no-recursion -C / -T /tmp/backupfiles) | \
  $COMPRESS_PROG | \
  ccrypt -k $KEYFILE -e | \
  catsync hcoop.backup.tar$COMPRESS_EXT.aescrypt
du -chs hcoop.backup.tar$COMPRESS_EXT.aescrypt

vos listvol deleuze | \
  tail -n +2 | \
  head -n -3 | \
  cut -b1-34 | \
  grep -v "\.backup .*$" | \
  grep -v "\.readonly .*$" | \
  cat > volumes

echo backing up databases
tar -C /var/backups/databases/ -cf - . | \
  $COMPRESS_PROG | \
  ccrypt -k $KEYFILE -e \
  | catsync databases.tar$COMPRESS_EXT.aescrypt

for A in `cat volumes | grep -v not-backed-up`
do \
  echo "dumping afs volume $A..."
  time (cd /; sudo vos dump -id $A -localauth -clone) | \
       $COMPRESS_PROG | \
       ccrypt -k $KEYFILE -e | \
       catsync $A.dump$COMPRESS_EXT.aescrypt
done