backup revision bacf95fddc53ab3107f380c4c816fa8072358bd9
# Description
# This backup script creates and restores backups of ontohub data. It includes:
# * bare git repositories (data/repositories)
# * named symlinks to git repositories (data/git_daemon)
# * the postgres database
#
# Usage
# First note: Run this as the ontohub user, *not* as root.
# To create a backup, run this script with the argument `create`:
# Then a backup named with the current date and time is created in the
# backup directory (see below).
#
# To restore a backup, run this script with the argument `restore <backup name>`
# Then the selected backup is fully restored
#
# Backup directory
# For development machines, the backup directory is:
# And for production machines, the backup directory is:
#
# Super user privileges
# To create and restore, we need root privileges. Otherwise file modes are not
# preserved. This script will call `sudo` when needed and inform you about the
# reason for calling `sudo`. If you don't allow sudo, a backup will be created
# or restored anyway, but the file modes and ownership are not preserved.
# Then, you need to adjust them manually.
#
# Maintenance mode
# While backing up and restoring the data, the maintenance mode is activated.
# This way we guarantee data consistency of the backup.
require 'tmpdir.rb'
require 'fileutils'
require 'pathname'
require 'open3'
# Amount of backups that have to be there at least
BACKUPS_COUNT = 30
# Backups are kept for at least 365 days
DATA_DIRS = %w(data/repositories data/git_daemon)
verbose: false, dry_run: true, sql_dump_as_postgres_user: false)
end
puts 'Creating backup...'
# We needed to create the directory for the script to continue later on.
puts "Created backup in #{backup_instance_dir}"
end
puts "Restored backup from #{backup_instance_dir}"
end
if !Dir.exists?(backup_root)
return
end
puts "removing old backup: #{dir}"
end
end
end
end
# Create directory even in dry run to let the script continue.
end
puts 'Creating SQL dump...'
end
end
puts 'Creating repository archive...'
end
end
"Error: Backup '#{backup_name}' does not exist in #{backup_root}.")
exit
end
end
'Restoring SQL dump...'
end
end
puts 'Restoring repository archive...'
rescue => e
puts <<-MSG
An error occured while restoring the repositories:
#{e.message}
You can find the pre-restore repositories at #{tmpdir}
Do something about it.
MSG
raise e
end
end
end
def move_data_dirs_to_tmpdir(tmpdir)
puts "FileUtils.mv(#{DATA_DIRS}, #{tmpdir})" if verbose
FileUtils.mv(DATA_DIRS, tmpdir) unless dry_run
rescue Errno::EACCES
puts <<-MSG
As the current user I have no access to move the repository data
directories #{DATA_DIRS.join(' ')} to a temporary directory #{tmpdir}.
This is used as a backup for the case of an error while restoring.
To continue, I try the command again using sudo.
MSG
exec('sudo', 'mv', *DATA_DIRS.map(&:to_s), tmpdir)
end
def extract_archive
archive_file = backup_instance_dir.join(REPOSITORY_FILE)
puts <<-MSG
Super user privileges are needed to reset the file permissions as
they were before the backup. If you refuse to enter the password
(Ctl-C) or enter a wrong password, only the permissions will not be
restored and all restored files will belong to the current user/group.
MSG
try_as_sudo_with_fallback('tar', verbose ? '-v' : '', '-xf',
archive_file.to_s, *DATA_DIRS.map(&:to_s))
end
def remove_tmpdir(tmpdir)
puts "FileUtils.remove_entry(#{tmpdir})" if verbose
FileUtils.remove_entry(tmpdir) # even do this in dry run
rescue Errno::EACCES
puts <<-MSG
As the current user I have no access to remove the temporary
directory #{tmpdir}.
To continue, I try the command again using sudo.
MSG
exec('sudo', 'rm', '-r', tmpdir)
end
def enable_maintenance_mode
puts 'Enabling maintenance mode...'
if File.exist?(maintenance_file)
$stderr.puts 'Maintenance mode was already enabled.'
$stderr.puts "Please check the file #{maintenance_file}"
$stderr.puts 'Aborting.'
exit
end
puts "FileUtils.touch #{maintenance_file}" if verbose
FileUtils.touch maintenance_file unless dry_run
end
def disable_maintenance_mode
puts 'Disabling maintenance mode...'
puts "FileUtils.rm #{maintenance_file}" if verbose
FileUtils.rm maintenance_file unless dry_run
end
def exec(*args)
puts "[executing next command in #{Dir.getwd}]" if verbose
out = args.join(' ')
puts out if verbose
system(*args) unless dry_run
end
def try_as_sudo_with_fallback(*args)
_out, _err, exit_code = exec('sudo', *args)
unless exit_code.success?
sudo_not_given_fallback(*args) # Wrong sudo password
end
rescue Exception => e
raise e unless e.is_a?(Interrupt) # Ctrl-C when asked for password
sudo_not_given_fallback(*args)
end
def sudo_not_given_fallback(*args)
puts 'Super user privileges not granted. Trying as normal user.'
exec(*args)
end
def maintenance_file
data_root.join(MAINTENANCE_FILE)
end
def pg_user_switch
sql_dump_as_postgres_user ? %w(-U postgres) : []
end
def self.backup_dirs_allowed_to_delete(entries)
entries.reject{ |entry| %w(. ..).include?(entry) }[0..-(BACKUPS_COUNT+1)]
end
end
end
def data_root(rails_root)
File.realpath(rails_root.join('data'))
end
def on_development_system?(rails_root)
!File.symlink?(rails_root.join('data'))
end
# Don't allow this to be run as the root user.
if ENV['USER'] == 'root'
puts 'Running this script as the root user is disabled.'
puts 'Please run it as a normal user that has sudo privileges, e.g. ontohub.'
exit
end
# We assume, this script runs in "RAILS_ROOT/script/".
RAILS_ROOT = Pathname.new(__FILE__).dirname.join('..')
BACKUP_ROOT_PRODUCTION = '/home/ontohub/ontohub_data_backup'
DATABASE =
if on_development_system?(RAILS_ROOT)
'ontohub_development'
else
'ontohub'
end
BACKUP_ROOT =
if on_development_system?(RAILS_ROOT)
RAILS_ROOT.join('tmp', 'backup')
else
File.realpath(BACKUP_ROOT_PRODUCTION)
end
backup = Backup::Backup.new(DATABASE, data_root(RAILS_ROOT), BACKUP_ROOT,
sql_dump_as_postgres_user: on_development_system?(RAILS_ROOT),
dry_run: false, verbose: true)
case ARGV.first
when 'create'
backup.create
when 'restore'
if ARGV.length == 1
$stderr.puts(
'To restore a backup, you need to specify one with the arguments')
$stderr.puts('"restore backup_name"')
exit
end
backup_name = ARGV[1]
backup.restore(backup_name)
when 'prune'
Backup::Backup.prune(BACKUP_ROOT)
else
$stderr.puts 'unknown or missing parameter'
$stderr.puts 'use parameter "create" or "restore <backup_name>" or "prune"'
exit
end