#!/usr/bin/python

# Joe G - 9/11/14

# LVM VM backup script whipped up in like 20 minutes. Accesses VM partitions inside LVs
# (named like vm-debianshiz) and copies files out of ones for linux and bsd with rsync
#
# TODO:
# - better error handling
# - create lv snapshot for each VM, fsck it, and go off that so this can run 
#   while the VMs are mounted & being used
# - suck less in general
# - make algorithms and flow more pythonic / less code
# - run rsync with ionice/nice
# - make this pep8/flake compatible (with exception for 2 spaces instead of 4)
# - use convenience shorter subprocess methods instead of just Popen

import subprocess
import time
import re
import os

FILESYSTEMS = ['ext4', 'jfs', 'ufs', 'ext3', 'ext2', 'btrfs', 'xfs']
MOUNTOPTIONS = ['ro', 'noatime']
FSOPTIONS = {'ufs': ['ufstype=ufs2']} # XXX: freebsd uses ufs2

TMPMOUNT = '/mnt/tmp'
BAKDEST = '/srv/media/backup/vm'

def getVMs():
  cmd = ['lvs']
  try:
    proc = subprocess.Popen(cmd, stdout=subprocess.PIPE, shell=False)
  except OSError:
    raise

  output = proc.communicate()[0]

  for line in output.split('\n')[1:]:
    m = re.search('^(vm\-\S+)\s+(\S+)', line.strip())
    if m:
      vm = m.group(1)[3:]
      yield (vm, '/dev/{1}/{0}'.format(*m.groups()))

def getPartitions(lv):
  parts = []

  # XXX: Get list of partitions inside LV
  cmd = ['kpartx', '-l', lv]

  try:
    proc = subprocess.Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE, shell=False)
  except OSError:
    raise

  output = proc.communicate()[0]

  for line in output.split('\n'):
    try:
      parts.append(line.strip().split()[0])
    except IndexError:
      pass

  if not len(parts):
    return
  
  # XXX: Make them accessible under /dev/mapper
  cmd = ['kpartx', '-a', lv]

  try:
    proc = subprocess.Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE, shell=False)
  except OSError:
    raise

  proc.wait()

  if proc.returncode != 0:
    cleanupPartitions(lv)
    return

  # XXX: Use blkid to get the filesystem of each partition
  cmd = ['blkid'] + ['/dev/mapper/{0}'.format(name) for name in parts]

  try:
    proc = subprocess.Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE, shell=False)
  except OSError:
    raise

  output = proc.communicate()[0]

  if proc.returncode != 0:
    cleanupPartitions(lv)
    return

  # TODO: make following code smaller
  for line in output.split('\n'):
    halves = line.split(':')
    path = halves[0]

    if not len(path) or len(halves) != 2:
      continue

    options = {}

    for pair in halves[1].split():
      parts = pair.replace('"', '').split('=')

      try:
        options[parts[0]] = parts[1]
      except IndexError:
        pass

    try:
      yield (path, options['TYPE'])
    except KeyError:
      pass

def cleanupPartitions(lv):

  # TODO: make this give up after 3 tries
  while True:

    time.sleep(1)

    cmd = ['kpartx', '-d', lv]

    try:
      proc = subprocess.Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE, shell=False)
    except OSError:
      raise

    out = proc.communicate()[0]

    if proc.returncode == 0:
      break

    print 'Trying again to unbind {0} {1} "{2}"..'.format(lv, proc.returncode, out)

def mount(src, dest, fs):
  options = MOUNTOPTIONS

  if fs in FSOPTIONS:
    options += FSOPTIONS[fs]

  cmd = ['mount', '-o', ','.join(options), src, dest]

  try:
    proc = subprocess.Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE, shell=False)
  except OSError:
    raise

  proc.wait()

  if proc.returncode != 0:
    raise Exception('Failed mounting')

def umount(path):

  # TODO: make this try 3 times, and run fuser -mvk $path on 
  # subsequent runs to kill open handles
  cmd = ['umount', path]

  try:
    proc = subprocess.Popen(cmd, shell=False)
  except OSError:
    raise

  out = proc.communicate()[0]

  if proc.returncode != 0:
    raise Exception('Failed umounting: {0}'.format(out))

def sync(src, dest):
  # TODO: make rsync exclude /usr/ports on ufs and the cache dirs for 
  # yum/apt-get on linux FSs. 
  
  if not src.endswith('/'):
    src += '/'

  if not dest.endswith('/'):
    dest += '/'

  cmd = ['rsync', '-avP', src, dest]

  try:

    try:
      proc = subprocess.Popen(cmd, shell=False)
    except OSError:
      raise

    proc.wait()

    if proc.returncode == 0:
      raise Exception('Failed syncing')

  except KeyboardInterrupt:
    print 'asshole'

for vm, lv in getVMs():
  for path, fs in getPartitions(lv):
    if fs not in FILESYSTEMS:
      continue

    tmpdest = TMPMOUNT
    bakroot = os.path.join(BAKDEST, vm, path[-1])

    if not os.path.exists(bakroot):
      os.makedirs(bakroot, 0700)

    try:
      mount(path, tmpdest, fs)
    except Exception as e:
      print e
      umount(tmpdest)
      continue

    try:
      sync(tmpdest, bakroot)
    except Exception as e:
      print e
      
    umount(tmpdest)

  cleanupPartitions(lv)