Try not backing up spamd data, in case that pushes us over the edge.

[hcoop/scripts.git] / hcoop-backup

#!/bin/bash -e

#
# it is dangerous to remove the "-e" above; please don't do that.
#

#
# run this script as root, on deleuze
#

exec >& /var/log/backup-to-megacz.com-log

PATH=$PATH:/bin:/usr/bin:/sbin:/usr/sbin
#COMPRESS_EXT=.bz2
#COMPRESS_PROG=bzip2
COMPRESS_EXT=.gz
COMPRESS_PROG=gzip

IFS=$'\n'

KEYFILE=/etc/backup-encryption-key
BACKUPDIR=/afs/megacz.com/private/hcoop-backup/

cd $BACKUPDIR
find * -prune -ctime +2 -exec rm -rf {} \; || true

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

groups
echo 'I am in:'
pwd
echo

echo building package lists...
dpkg-query -W -f='${Package}\n' > packages
(cd /; find / /usr/ /usr/local/ /var/ -xdev)   | sort | uniq > allfiles
dpkg-query -W -f='${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/local/lib/spamd | \
 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/$' | \
 grep -v ^/usr/src | \
 grep -v '^/usr/.*\.pyc' | \
 grep -v '^/usr/.*\.elc' | \
 grep -v '^/usr/bin/perldoc\.stub$' | \
 grep -v '^/usr/bin/.*\.notslocate$' | \
 grep -v '^/usr/lib/courier/.*\.rand$' | \
 grep -v '^/usr/lib/gconv/gconv-modules\.cache$' | \
 grep -v '^/usr/lib/graphviz/config$' | \
 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 ^/usr/share/perl5/IkiWiki/Plugin | \
 grep -v ^/media | \
 grep -v ^/vmlinuz | \
 grep -v ^/vmlinuz.old | \
 grep -v '^/sbin/[a-z\-]*\.modutils$' | \
 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/ | \
 xargs -I{} -d\\n -- bash -c "test -L '{}' || echo '{}'" | \
 cat > complain

tar clpf - --ignore-failed-read --no-recursion -C / -T backupfiles | \
  $COMPRESS_PROG | \
  ccrypt -k $KEYFILE -e | \
  catsync hcoop.backup.tar$COMPRESS_EXT.aescrypt

vos listvol deleuze | \
  tail -n +2 | \
  head -n -3 | \
  cut -b1-34 | \
  grep -v "\.backup .*$" | \
  grep -v "\.readonly .*$" | \
  sed 's_^ .*__' | \
  sed 's_ .*$__' | \
  grep '[A-Za-z]' | \
  cat > volumes

cat volumes | \
  grep -v not-backed-up | \
  xargs -I{} -d\\n -- \
  bash -c \
    "vos dump -id {} -localauth -clone | $COMPRESS_PROG | ccrypt -k $KEYFILE -e | catsync {}.dump$COMPRESS_EXT.aescrypt"

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

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 \
  || true

echo done