xwiki-backup-s3

#!/bin/bash

# fail immediately on error
set -e
set -o pipefail
#set -x    # debug mode

PROGNAME=$(basename $0)
promdata_dir=/var/lib/prometheus # no slash on end
backup_filename=xwiki-backup.tar.gz
s3_backup_location="s3://MYBUCKET/xwiki/$backup_filename"
pghost=RDSPREFIX.ap-southeast-2.rds.amazonaws.com
pgport=5432
pguser=xwiki
pgdb=xwikidb
pgpass=MYPASS
xwiki_cfg_files=(hibernate.cfg.xml
                 xwiki.cfg
                 xwiki.properties
                 classes/logback.xml
                )

usage () {
cat <<EOF
Usage:
    $PROGNAME go

$PROGNAME is used to back up the Xwiki installation, as per the guidelines
at https://www.xwiki.org/xwiki/bin/view/Documentation/AdminGuide/Backup

It takes a psql dump, config file copy, and 'permanent dir' copy, tars them up,
and deposits them on s3.

This script is designed to be run as a cronjob, and should be run as the Tomcat user, in order to access the xwiki 'permanent' data directory.

Requirements:
- aws cli tool
- pg_dump 10.4+
- s3 perms
  - this script assumes role-based perms, so no keys stored here

$PROGNAME is intended to be run as a cronjob.

The S3 bucket should have a lifecycle rule on this location to expire old
backups after X days, and older versions of the backup will simply be earlier
versions of the uploaded backup file.

Restoring the backups should simply be running pg_restore, and copying back the tarred files. 'Should'.

EOF
exit 0
}

if [ -z "$1" ]; then usage; fi

cleanup(){
    log "Removing temporary directory..."
    rm -r $TMPDIR
    log "Done."
}

log(){
    echo "$(date -Is)> $@"
}

trap cleanup exit # on any exit

log "Backing up LKG xwiki data..."

TMPDIR="$(mktemp -d)"

mkdir -p "$TMPDIR"/{psql,config/classes,config,data}

cd "$TMPDIR"

log "Dumping postgres database..."
PGPASSWORD="$pgpass" pg_dump \
    --host "$pghost" \
    --port "$pgport" \
    --username "$pguser" \
    --dbname "$pgdb" \
    --format c \
    --file psql/xwikidb.pgdump

log "Fetching config files..."
# cluster info in /observation/ is not relevant for us
for file in ${xwiki_cfg_files[@]}; do
    cp "/etc/xwiki/$file" "config/$file"
done

log "Fetching 'permanent' data..."
rsync -a /var/lib/xwiki/data/ data/

cat > README.txt <<EOF
Backup taken: $(date)

This is a backup of the LKG Xwiki installation, and comes in three parts as per
https://www.xwiki.org/xwiki/bin/view/Documentation/AdminGuide/Backup

1. psql dump - if restoring to a new pghost, the jdbc link in hibernate.xml
   will need to be updated
2. config files - these get copied into /etc/xwiki/
3. xwiki 'permanent' data dir - this gets copied to /var/lib/xwiki/data

The xwiki install is installed via Debian packages, from the official XWiki apt
repo. Other forms of installation will have different target locations to the
ababove.
EOF

log "Tarring up backup files..."
tar zcf "$backup_filename" *

log "Uploading to (versioned) s3 location ($s3_backup_location)..."
/usr/local/bin/aws s3 cp "$backup_filename" "$s3_backup_location"

# cleanup happens on exit