clonedb.py 19 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545
  1. """
  2. Script de clonage des bases de données MySql
  3. (requiert python 3.6+)
  4. > Configuration: settings.yml
  5. Usage:
  6. clonedb.py [-v] [-y] [<opname>...]
  7. clonedb.py (-h | --help)
  8. clonedb.py --version
  9. Options:
  10. -y, --yes Do not ask for confirmation
  11. -h --help Show this screen.
  12. --version Show version.
  13. @author: olivier.massot, 05-2020
  14. """
  15. import logging
  16. import re
  17. from subprocess import Popen, PIPE, CalledProcessError
  18. import sys
  19. import pymysql
  20. import yaml
  21. from docopt import docopt
  22. from path import Path
  23. from core import logging_
  24. from core.docker import resolve_docker_ip
  25. from core.locker import Lockfile
  26. from core.pipe_handler import PipeHandler
  27. from core.ssh import SshTunnel
  28. from core.prompt import ask_confirmation
  29. __VERSION__ = "0.2"
  30. HERE = Path(__file__).parent
  31. # Start logger
  32. LOG_DIR = HERE / 'log'
  33. LOG_DIR.mkdir_p()
  34. logger = logging.getLogger('clonedb')
  35. logging_.start("clonedb", filename=LOG_DIR / 'clonedb.log', replace=True)
  36. # FIX the default ascii encoding on some linux dockers...
  37. sys.stdout = open(sys.stdout.fileno(), mode='w', encoding='utf8', buffering=1)
  38. # Options
  39. SHOW_PROGRESSION = True
  40. LOG_PIPES_OUTPUT = False
  41. LOG_MYSQL_QUERIES = False
  42. MY_CNF_FILE = Path(HERE / "my.cnf")
  43. MAX_ALLOWED_PACKET = 1073741824
  44. # Utilities
  45. def load_settings():
  46. """ Load the settings from the 'settings.yml' file
  47. If there is no such file, the base settings.yml file is created
  48. """
  49. settings_file = HERE / 'settings.yml'
  50. if not settings_file.exists():
  51. Path(HERE / 'settings.yml.dist').copy(HERE / 'settings.yml')
  52. with open(settings_file, 'r') as f:
  53. return yaml.load(f, Loader=yaml.FullLoader)
  54. def _print(msg, end=False):
  55. msg = msg.ljust(80)
  56. print(f'\r{msg}', end='' if not end else '\n', flush=True)
  57. class MysqldumpHandler(PipeHandler):
  58. """ Handle and process the stdout / stderr output from a mysqldump process
  59. """
  60. _rx_prog = re.compile(r'Retrieving table structure for table (\w+)')
  61. _log_all = LOG_PIPES_OUTPUT
  62. def __init__(self, logger_name, level, total_prog):
  63. super().__init__(logger_name, level)
  64. self.total_prog = total_prog
  65. self.prog = 0
  66. self._last_logged = ""
  67. def process(self, line):
  68. """ Process the last line that was read
  69. """
  70. line = line.strip('\n')
  71. if SHOW_PROGRESSION:
  72. match = self._rx_prog.search(line)
  73. if match:
  74. self.log_new_table(match.group(1), "dumping")
  75. if self._log_all:
  76. logger.debug(line)
  77. def log_new_table(self, tname, action_name=""):
  78. if tname == self._last_logged:
  79. return
  80. self.prog += 1
  81. logger.debug('... %s %s', action_name, tname)
  82. _print(f'{action_name} `{tname}` [{self.prog} / {self.total_prog}]')
  83. self._last_logged = tname
  84. def log_end(self):
  85. _print(f'\r-- done --', end=True)
  86. def close(self):
  87. """ Close the write end of the pipe.
  88. """
  89. super().close()
  90. class MysqlHandler(MysqldumpHandler):
  91. """ Handle and process the stdout / stderr output from a mysql process
  92. """
  93. _rx_prog = re.compile(r'^((?:CREATE TABLE )|(?:INSERT INTO ))`(\w+)`')
  94. _log_all = LOG_PIPES_OUTPUT
  95. _action_name = "restoring"
  96. def process(self, line):
  97. """ Process the last line that was read
  98. """
  99. line = line.strip('\n')
  100. if SHOW_PROGRESSION:
  101. match = self._rx_prog.search(line)
  102. if match:
  103. action_name = "restoring {}".format('structure of'
  104. if 'CREATE' in match.group(1)
  105. else 'data of')
  106. self.log_new_table(match.group(2), action_name)
  107. if self._log_all:
  108. logger.debug(line)
  109. class MySqlServer:
  110. """ A server hosting a Mysql instance
  111. """
  112. def __init__(self, host, port, username, password, description="", ssh_tunnel=None):
  113. self.host = host
  114. self.port = port
  115. self.username = username
  116. self.password = password
  117. self.description = description[:30]
  118. self.ssh_tunnel = ssh_tunnel
  119. self.cnn = None
  120. self.active_db = ""
  121. def __repr__(self):
  122. s = f"{self.host}:{self.port} as {self.username}"
  123. if self.description:
  124. s = f"{self.description} ({s})"
  125. return s
  126. def connect(self, autocommit=True):
  127. """ Establish the connection to the Mysql server
  128. @see https://pymysql.readthedocs.io/en/latest/modules/connections.html
  129. """
  130. if self.ssh_tunnel:
  131. self.ssh_tunnel.start()
  132. host, port = self.ssh_tunnel.LOCAL_ADRESS
  133. else:
  134. host, port = self.host, self.port
  135. self.cnn = pymysql.connect(host=host,
  136. port=port,
  137. user=self.username,
  138. password=self.password,
  139. autocommit=autocommit,
  140. max_allowed_packet="32M",
  141. )
  142. if not self.cnn.open:
  143. raise RuntimeError(f'Unable to connect to {self}')
  144. return self.cnn
  145. def set_active_db(self, dbname):
  146. """ set the active database
  147. """
  148. self.cnn.select_db(dbname)
  149. self.active_db = dbname
  150. def close(self):
  151. """ Close the connection to the database
  152. and the ssh tunnel if one is opened
  153. """
  154. if self.cnn:
  155. self.cnn.close()
  156. if self.ssh_tunnel:
  157. self.ssh_tunnel.stop()
  158. logger.debug(f'{self} - connection closed')
  159. def exec_query(self, sql):
  160. """ Execute the sql code and return the resulting cursor
  161. @see https://pymysql.readthedocs.io/en/latest/modules/cursors.html
  162. """
  163. self.cnn.ping(reconnect=True)
  164. cursor = self.cnn.cursor()
  165. if LOG_MYSQL_QUERIES:
  166. logger.debug(sql)
  167. cursor.execute(sql)
  168. return cursor
  169. def db_exists(self, dbname):
  170. """ Return True if the database exists
  171. """
  172. cursor = self.exec_query(f"""SELECT SCHEMA_NAME
  173. FROM INFORMATION_SCHEMA.SCHEMATA
  174. WHERE SCHEMA_NAME = '{dbname}'""")
  175. row = cursor.fetchone()
  176. return row is not None
  177. def list_tables(self, dbname=""):
  178. """ Return a list of tables (but not views!)
  179. for either the currently selected database,
  180. or the one given as a parameter"""
  181. cursor = self.exec_query(
  182. "SHOW FULL TABLES{} WHERE Table_type='BASE TABLE';".format(f" FROM {dbname}" if dbname else ""))
  183. return (row[0] for row in cursor.fetchall())
  184. def list_views(self, dbname=""):
  185. """ Return a list of views
  186. for either the currently selected database,
  187. or the one given as a parameter"""
  188. cursor = self.exec_query(
  189. "SHOW FULL TABLES{} WHERE Table_type='VIEW';".format(f" FROM {dbname}" if dbname else ""))
  190. return (row[0] for row in cursor.fetchall())
  191. def get_view_definition(self, view_name, set_definer=""):
  192. """ Return the SQL create statement for the view
  193. If 'set_definer' is not empty, the username in the 'SET DEFINER' part
  194. of the create statement is replace by the one given
  195. """
  196. cursor = self.exec_query(f"show create view {view_name}")
  197. definition = cursor.fetchone()[1]
  198. if set_definer:
  199. # force a new definer
  200. definition = re.sub(r'DEFINER=`\w+`@`[\w\-.]+`',
  201. f"DEFINER=`{set_definer}`@`\1`",
  202. definition)
  203. return definition
  204. # Operation status
  205. UNKNOWN = 0
  206. SUCCESS = 1
  207. FAILURE = 2
  208. # Behaviors for the tables cloning
  209. IGNORE = 0
  210. STRUCTURE_ONLY = 1
  211. STRUCTURE_AND_DATA = 2 # -> default behavior
  212. class CloningOperation:
  213. """ A database cloning operation between two Mysql servers
  214. """
  215. def __init__(self, name, dbname, from_server, to_server, is_default=True, ignore_tables=None, structure_only=None,
  216. filter_tables=None, ignore_views=None, compress=True):
  217. self.name = name
  218. self.dbname = dbname
  219. self.from_server = from_server
  220. self.to_server = to_server
  221. self.is_default = is_default
  222. self.compress = compress
  223. self.ignore_tables = [re.compile(r) for r in ignore_tables] if ignore_tables else []
  224. self.structure_only = [re.compile(r) for r in structure_only] if structure_only else []
  225. self.filter_tables = [re.compile(r) for r in filter_tables] if filter_tables else []
  226. self.ignore_views = [re.compile(r) for r in ignore_views] if ignore_views else []
  227. self.status = UNKNOWN
  228. def __repr__(self):
  229. return f"Cloning {self.dbname} from {self.from_server} to {self.to_server}"
  230. def _build_dump_command(self, dump_options=None, tables=None):
  231. """ Build a mysqldump command line and return it as a
  232. ready-to-consume list for Popen
  233. @see https://dev.mysql.com/doc/refman/5.7/en/mysqldump.html#mysqldump-option-summary
  234. """
  235. tables = tables or []
  236. dump_options = dump_options or []
  237. base_cmd = ["mysqldump",
  238. "--single-transaction",
  239. "-u", self.from_server.username,
  240. f"--password={self.from_server.password}",
  241. f"--max-allowed-packet={MAX_ALLOWED_PACKET}",
  242. "--skip-add-drop-table",
  243. "--skip-add-locks",
  244. "--skip-comments",
  245. ]
  246. if self.compress:
  247. base_cmd.append("--compress")
  248. if SHOW_PROGRESSION:
  249. base_cmd.append("--verbose")
  250. if self.from_server.ssh_tunnel:
  251. host, port = self.from_server.ssh_tunnel.LOCAL_ADRESS
  252. base_cmd += ["--host", host,
  253. "--port", str(port)]
  254. return base_cmd + dump_options + [self.dbname] + tables
  255. def _build_restore_command(self):
  256. """ Build a mysql command line and return it as a
  257. ready-to-consume list for Popen
  258. @see https://dev.mysql.com/doc/refman/8.0/en/mysql-command-options.html#option_mysql_quick
  259. """
  260. init_command = f"set global max_allowed_packet={MAX_ALLOWED_PACKET};" \
  261. "set global wait_timeout=28800;" \
  262. "set global interactive_timeout=28800;"
  263. cmd = ["mysql",
  264. "-h", self.to_server.host,
  265. "-P", str(self.to_server.port),
  266. "-u", self.to_server.username,
  267. f"--password={self.to_server.password}",
  268. f"--init-command={init_command}",
  269. "--reconnect",
  270. "--quick",
  271. "--unbuffered",
  272. "--wait",
  273. "--verbose",
  274. "-D", self.dbname
  275. ]
  276. # if LOG_PIPES_OUTPUT:
  277. # cmd.append("--verbose")
  278. if self.compress:
  279. cmd.append("--compress")
  280. return cmd
  281. @staticmethod
  282. def _run_piped_processes(dump_cmd, restore_cmd, tbl_count):
  283. """ Run the dump and the restore commands by piping them
  284. The output of the mysqldump process is piped into the input of the mysql one
  285. """
  286. logger.debug(">>> Dump command: %s", " ".join(map(str, dump_cmd)))
  287. logger.debug(">>> Piped into: %s", " ".join(map(str, restore_cmd)))
  288. mysqldump_handler = MysqldumpHandler(logger.name, logging.INFO, tbl_count)
  289. mysql_handler = MysqlHandler(logger.name, logging.INFO, tbl_count)
  290. try:
  291. # noinspection PyTypeChecker
  292. with Popen(restore_cmd, stdin=PIPE, stdout=mysql_handler, stderr=mysql_handler) as mysql:
  293. # noinspection PyTypeChecker
  294. with Popen(dump_cmd, stdout=PIPE, stderr=mysqldump_handler) as mysqldump:
  295. mysql.stdin.write(mysqldump.stdout.read())
  296. if mysqldump.returncode:
  297. raise RuntimeError('mysqldump returned a non zero code')
  298. if mysql.returncode:
  299. raise RuntimeError('mysql returned a non zero code')
  300. mysql_handler.log_end()
  301. except (OSError, RuntimeError, CalledProcessError) as e:
  302. logger.error("Execution failed: %s", e)
  303. raise RuntimeError(f"An error happened at runtime: {e}")
  304. finally:
  305. mysqldump_handler.close()
  306. mysql_handler.close()
  307. def run(self):
  308. """ Run the cloning op
  309. """
  310. logger.info(f"*** Cloning {self.dbname} ***")
  311. logger.info(f"> From {self.from_server}")
  312. logger.info(f"> To {self.to_server}")
  313. try:
  314. self.from_server.connect()
  315. self.from_server.set_active_db(self.dbname)
  316. logger.debug('Connected to %s', self.from_server)
  317. self.to_server.connect()
  318. logger.debug('Connected to %s', self.to_server)
  319. tables = {}
  320. for tname in self.from_server.list_tables():
  321. if any(rx.match(tname) for rx in self.ignore_tables):
  322. tables[tname] = IGNORE
  323. elif self.filter_tables and not any(rx.match(tname) for rx in self.filter_tables):
  324. tables[tname] = IGNORE
  325. elif any(rx.match(tname) for rx in self.structure_only):
  326. tables[tname] = STRUCTURE_ONLY
  327. else:
  328. tables[tname] = STRUCTURE_AND_DATA
  329. restore_cmd = self._build_restore_command()
  330. # Dump structure: --single-transaction --no-data --routines {dbname} tbname1 tname2 ...
  331. dump_structure_for = [t for t, s in tables.items() if s != IGNORE]
  332. dump_structure_cmd = self._build_dump_command(["--no-data", "--routines"],
  333. dump_structure_for)
  334. # Dump data: --no-create-info --skip-triggers {dbname} tbname1 tname2 ...
  335. dump_data_for = [t for t, s in tables.items() if s == STRUCTURE_AND_DATA]
  336. dump_data_cmd = self._build_dump_command(["--no-create-info", "--skip-triggers"],
  337. dump_data_for)
  338. if tables and not dump_structure_for and not dump_data_for:
  339. logging.warning('No table will be cloned')
  340. # Recreate the target DB
  341. logger.info("(Re)create the database")
  342. self.to_server.exec_query(f"DROP DATABASE IF EXISTS `{self.dbname}`;")
  343. self.to_server.exec_query(f"CREATE SCHEMA `{self.dbname}`;")
  344. self.to_server.set_active_db(self.dbname)
  345. # Run mysqldump
  346. try:
  347. if dump_structure_for:
  348. logger.info(f"Cloning structure for {len(dump_structure_for)} tables (on {len(tables)})...")
  349. self._run_piped_processes(dump_structure_cmd, restore_cmd, len(dump_structure_for))
  350. if dump_data_for:
  351. logger.info(f"Cloning data for {len(dump_data_for)} tables (on {len(tables)})...")
  352. self._run_piped_processes(dump_data_cmd, restore_cmd, len(dump_data_for))
  353. logger.info(f"Cloning views...")
  354. self.from_server.set_active_db(self.dbname)
  355. for v in self.from_server.list_views(self.dbname):
  356. if any(rx.match(v) for rx in self.ignore_views):
  357. continue
  358. logger.debug('* cloning view %s', v)
  359. definition = self.from_server.get_view_definition(v, self.to_server.username)
  360. try:
  361. self.to_server.exec_query(definition)
  362. except (pymysql.err.ProgrammingError, pymysql.err.InternalError) as e:
  363. logger.error('Unable to create the internal view %s: %s', v, e)
  364. self.status = SUCCESS
  365. logger.info("> the database was successfully cloned")
  366. except RuntimeError:
  367. self.status = FAILURE
  368. logger.error("<!> An error happened while cloning the '%s' database", self.dbname)
  369. finally:
  370. self.from_server.close()
  371. self.to_server.close()
  372. def main(settings, arguments):
  373. prompt = not arguments["--yes"]
  374. logger.info("Start db cloning utility...")
  375. logger.debug(f"Settings: %s", str(settings).replace('\r', '').replace('\n', ''))
  376. logger.debug(f"Arguments: %s", str(arguments).replace('\r', '').replace('\n', ''))
  377. # Load the servers' configuration
  378. servers = {}
  379. for server_name, server_settings in settings['servers'].items():
  380. hostname = server_settings['host']
  381. match = re.search(r"^docker:(\w+)$", hostname)
  382. if match:
  383. logger.debug("resolve IP for docker %s", match.group(1))
  384. ip = resolve_docker_ip(match.group(1))
  385. logger.debug("substitute '%s' to '%s' as hostname", ip, hostname)
  386. hostname = ip
  387. if 'ssh' in server_settings:
  388. ssh_tunnel = SshTunnel(hostname, server_settings['mysql']['port'], **server_settings['ssh'])
  389. else:
  390. ssh_tunnel = None
  391. server = MySqlServer(hostname,
  392. **server_settings['mysql'],
  393. description=server_settings['description'],
  394. ssh_tunnel=ssh_tunnel)
  395. servers[server_name] = server
  396. # Load the cloning ops' configuration
  397. ops = {}
  398. for name, args in settings['operations'].items():
  399. dbname = args['dbname']
  400. from_server = servers[args['from_server']]
  401. to_server = servers[args['to_server']]
  402. kwargs = {k: v for k, v in args.items() if k not in ('dbname', 'from_server', 'to_server')}
  403. op = CloningOperation(name, dbname, from_server, to_server, **kwargs)
  404. ops[name] = op
  405. # Operations to launch
  406. if arguments.get('<opname>', None):
  407. selected_ops = []
  408. for opname in arguments['<opname>']:
  409. try:
  410. selected_ops.append(ops[opname])
  411. except KeyError:
  412. logger.error('No operation found with name %s', opname)
  413. else:
  414. selected_ops = [op for op in ops.values() if op.is_default]
  415. if not selected_ops:
  416. raise RuntimeError('No operation to launch')
  417. # Ask for confirmation (except if '--yes' is in arguments)
  418. if prompt:
  419. logger.debug('Ask for confirmation...')
  420. msg = "The following operations will be launched:\n{}\n" \
  421. "WARNING: the existing local databases will be replaced" \
  422. "".format("\n".join(f"* {op}" for op in selected_ops))
  423. if not ask_confirmation(msg):
  424. logger.info("-- Operation cancelled by user --")
  425. return
  426. logger.debug('> User confirmed')
  427. # Run the cloning operations
  428. for op in selected_ops:
  429. op.run()
  430. failures = [op.name for op in selected_ops if op.status == FAILURE]
  431. if failures:
  432. logger.error("WARNING! the following operations failed: %s", ', '.join(failures))
  433. if __name__ == '__main__':
  434. # load settings from settings.yml file
  435. settings = load_settings()
  436. # parse CLI arguments
  437. arguments = docopt(__doc__, help=__doc__, version=__VERSION__)
  438. with Lockfile(path=HERE / '.clonedb.lock',
  439. on_error=lambda: logger.critical("A cloning process is already running, please wait...")):
  440. main(settings, arguments)