mongo-cluster-backup.py

import commands
import pymongo
import time

from pymongo import Connection
from boto.ec2.connection import EC2Connection

host = '127.0.0.1'
timestamp = time.strftime("%Y-%m-%d-%H-%M")

CONFIGS_HOST = 'mongoc-server'
CONFIGS_PORT = 27019

shards = [
  {
    'name': 'shard-0',
    'port': 27017,
    'volume': 'vol-1212'
  },
  {
    'name': 'shard-1',
    'port': 27018,
    'volume': 'vol-23323'
  },
  {
    'name': 'shard-2',
    'port': 27019,
    'volume': 'vol-3434'
  }
]

ec2 = EC2Connection('key', 'secret')

# create a connection to the mongo-configs server and get the current
# balancer status
db = Connection(CONFIGS_HOST, CONFIGS_PORT)['config']
pre_backup_balancer_status = db.settings.find({'_id': 'balancer'}).next()

# stop the autobalancer
db.settings.update({'_id': 'balancer'}, {'$set': {'stopped': True}})

try:
  for shard in shards:
    print '%s: starting backup' % shard['name']

    mongodb = Connection(host, shard['port'], slave_okay=True)
    volume = ec2.get_all_volumes(shard['volume'])[0]

    print '%s: locking mongodb' % shard['name']

    mongodb.admin.command('fsync', lock=True)

    print '%s: initiating volume snapshot' % shard['name']

    description = "%s-%s" % (shard['name'], timestamp)

    try:
      print '%s: %s' % (shard['name'], volume.create_snapshot(description))
    except:
      print '%s: could not initiate snapshot' % shard['name']

    print '%s: unlocking mongodb' % shard['name']

    mongodb.admin['$cmd'].sys.unlock.find_one()

    print '%s: backup complete' % shard['name']

  # back up the mongo configs server
  status, output = commands.getstatusoutput(
    'mongodump -h %s:%d -d config -o /backup/mongoc/%s' \
    % (CONFIGS_HOST, CONFIGS_PORT, timestamp))

  if status: print 'error dumping config db: %s' % o
except:
  print 'oops, something bad happened during the shard backups :('

if pre_backup_balancer_status['stopped'] is False:
  db.settings.update({'_id': 'balancer'}, {'$set': {'stopped': False}})