synapse_port_db 30 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910911912913914915916917918919920921922923924925926927928929930931932933934935936937938939940941942943944945946947948949950951952953954955956957958959960961962963964965966967968969970971972973974975976
  1. #!/usr/bin/env python
  2. # -*- coding: utf-8 -*-
  3. # Copyright 2015, 2016 OpenMarket Ltd
  4. #
  5. # Licensed under the Apache License, Version 2.0 (the "License");
  6. # you may not use this file except in compliance with the License.
  7. # You may obtain a copy of the License at
  8. #
  9. # http://www.apache.org/licenses/LICENSE-2.0
  10. #
  11. # Unless required by applicable law or agreed to in writing, software
  12. # distributed under the License is distributed on an "AS IS" BASIS,
  13. # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
  14. # See the License for the specific language governing permissions and
  15. # limitations under the License.
  16. from twisted.internet import defer, reactor
  17. from twisted.enterprise import adbapi
  18. from synapse.storage._base import LoggingTransaction, SQLBaseStore
  19. from synapse.storage.engines import create_engine
  20. from synapse.storage.prepare_database import prepare_database
  21. import argparse
  22. import curses
  23. import logging
  24. import sys
  25. import time
  26. import traceback
  27. import yaml
  28. logger = logging.getLogger("synapse_port_db")
  29. BOOLEAN_COLUMNS = {
  30. "events": ["processed", "outlier", "contains_url"],
  31. "rooms": ["is_public"],
  32. "event_edges": ["is_state"],
  33. "presence_list": ["accepted"],
  34. "presence_stream": ["currently_active"],
  35. "public_room_list_stream": ["visibility"],
  36. "device_lists_outbound_pokes": ["sent"],
  37. "users_who_share_rooms": ["share_private"],
  38. "groups": ["is_public"],
  39. "group_rooms": ["is_public"],
  40. "group_users": ["is_public", "is_admin"],
  41. "group_summary_rooms": ["is_public"],
  42. "group_room_categories": ["is_public"],
  43. "group_summary_users": ["is_public"],
  44. "group_roles": ["is_public"],
  45. "local_group_membership": ["is_publicised", "is_admin"],
  46. }
  47. APPEND_ONLY_TABLES = [
  48. "event_content_hashes",
  49. "event_reference_hashes",
  50. "event_signatures",
  51. "event_edge_hashes",
  52. "events",
  53. "event_json",
  54. "state_events",
  55. "room_memberships",
  56. "feedback",
  57. "topics",
  58. "room_names",
  59. "rooms",
  60. "local_media_repository",
  61. "local_media_repository_thumbnails",
  62. "remote_media_cache",
  63. "remote_media_cache_thumbnails",
  64. "redactions",
  65. "event_edges",
  66. "event_auth",
  67. "received_transactions",
  68. "sent_transactions",
  69. "transaction_id_to_pdu",
  70. "users",
  71. "state_groups",
  72. "state_groups_state",
  73. "event_to_state_groups",
  74. "rejections",
  75. "event_search",
  76. "presence_stream",
  77. "push_rules_stream",
  78. "current_state_resets",
  79. "ex_outlier_stream",
  80. "cache_invalidation_stream",
  81. "public_room_list_stream",
  82. "state_group_edges",
  83. "stream_ordering_to_exterm",
  84. ]
  85. end_error_exec_info = None
  86. class Store(object):
  87. """This object is used to pull out some of the convenience API from the
  88. Storage layer.
  89. *All* database interactions should go through this object.
  90. """
  91. def __init__(self, db_pool, engine):
  92. self.db_pool = db_pool
  93. self.database_engine = engine
  94. _simple_insert_txn = SQLBaseStore.__dict__["_simple_insert_txn"]
  95. _simple_insert = SQLBaseStore.__dict__["_simple_insert"]
  96. _simple_select_onecol_txn = SQLBaseStore.__dict__["_simple_select_onecol_txn"]
  97. _simple_select_onecol = SQLBaseStore.__dict__["_simple_select_onecol"]
  98. _simple_select_one = SQLBaseStore.__dict__["_simple_select_one"]
  99. _simple_select_one_txn = SQLBaseStore.__dict__["_simple_select_one_txn"]
  100. _simple_select_one_onecol = SQLBaseStore.__dict__["_simple_select_one_onecol"]
  101. _simple_select_one_onecol_txn = SQLBaseStore.__dict__[
  102. "_simple_select_one_onecol_txn"
  103. ]
  104. _simple_update_one = SQLBaseStore.__dict__["_simple_update_one"]
  105. _simple_update_one_txn = SQLBaseStore.__dict__["_simple_update_one_txn"]
  106. _simple_update_txn = SQLBaseStore.__dict__["_simple_update_txn"]
  107. def runInteraction(self, desc, func, *args, **kwargs):
  108. def r(conn):
  109. try:
  110. i = 0
  111. N = 5
  112. while True:
  113. try:
  114. txn = conn.cursor()
  115. return func(
  116. LoggingTransaction(txn, desc, self.database_engine, [], []),
  117. *args, **kwargs
  118. )
  119. except self.database_engine.module.DatabaseError as e:
  120. if self.database_engine.is_deadlock(e):
  121. logger.warn("[TXN DEADLOCK] {%s} %d/%d", desc, i, N)
  122. if i < N:
  123. i += 1
  124. conn.rollback()
  125. continue
  126. raise
  127. except Exception as e:
  128. logger.debug("[TXN FAIL] {%s} %s", desc, e)
  129. raise
  130. return self.db_pool.runWithConnection(r)
  131. def execute(self, f, *args, **kwargs):
  132. return self.runInteraction(f.__name__, f, *args, **kwargs)
  133. def execute_sql(self, sql, *args):
  134. def r(txn):
  135. txn.execute(sql, args)
  136. return txn.fetchall()
  137. return self.runInteraction("execute_sql", r)
  138. def insert_many_txn(self, txn, table, headers, rows):
  139. sql = "INSERT INTO %s (%s) VALUES (%s)" % (
  140. table,
  141. ", ".join(k for k in headers),
  142. ", ".join("%s" for _ in headers)
  143. )
  144. try:
  145. txn.executemany(sql, rows)
  146. except:
  147. logger.exception(
  148. "Failed to insert: %s",
  149. table,
  150. )
  151. raise
  152. class Porter(object):
  153. def __init__(self, **kwargs):
  154. self.__dict__.update(kwargs)
  155. @defer.inlineCallbacks
  156. def setup_table(self, table):
  157. if table in APPEND_ONLY_TABLES:
  158. # It's safe to just carry on inserting.
  159. row = yield self.postgres_store._simple_select_one(
  160. table="port_from_sqlite3",
  161. keyvalues={"table_name": table},
  162. retcols=("forward_rowid", "backward_rowid"),
  163. allow_none=True,
  164. )
  165. total_to_port = None
  166. if row is None:
  167. if table == "sent_transactions":
  168. forward_chunk, already_ported, total_to_port = (
  169. yield self._setup_sent_transactions()
  170. )
  171. backward_chunk = 0
  172. else:
  173. yield self.postgres_store._simple_insert(
  174. table="port_from_sqlite3",
  175. values={
  176. "table_name": table,
  177. "forward_rowid": 1,
  178. "backward_rowid": 0,
  179. }
  180. )
  181. forward_chunk = 1
  182. backward_chunk = 0
  183. already_ported = 0
  184. else:
  185. forward_chunk = row["forward_rowid"]
  186. backward_chunk = row["backward_rowid"]
  187. if total_to_port is None:
  188. already_ported, total_to_port = yield self._get_total_count_to_port(
  189. table, forward_chunk, backward_chunk
  190. )
  191. else:
  192. def delete_all(txn):
  193. txn.execute(
  194. "DELETE FROM port_from_sqlite3 WHERE table_name = %s",
  195. (table,)
  196. )
  197. txn.execute("TRUNCATE %s CASCADE" % (table,))
  198. yield self.postgres_store.execute(delete_all)
  199. yield self.postgres_store._simple_insert(
  200. table="port_from_sqlite3",
  201. values={
  202. "table_name": table,
  203. "forward_rowid": 1,
  204. "backward_rowid": 0,
  205. }
  206. )
  207. forward_chunk = 1
  208. backward_chunk = 0
  209. already_ported, total_to_port = yield self._get_total_count_to_port(
  210. table, forward_chunk, backward_chunk
  211. )
  212. defer.returnValue(
  213. (table, already_ported, total_to_port, forward_chunk, backward_chunk)
  214. )
  215. @defer.inlineCallbacks
  216. def handle_table(self, table, postgres_size, table_size, forward_chunk,
  217. backward_chunk):
  218. if not table_size:
  219. return
  220. self.progress.add_table(table, postgres_size, table_size)
  221. if table == "event_search":
  222. yield self.handle_search_table(
  223. postgres_size, table_size, forward_chunk, backward_chunk
  224. )
  225. return
  226. if table in (
  227. "user_directory", "user_directory_search", "users_who_share_rooms",
  228. "users_in_pubic_room",
  229. ):
  230. # We don't port these tables, as they're a faff and we can regenreate
  231. # them anyway.
  232. self.progress.update(table, table_size) # Mark table as done
  233. return
  234. if table == "user_directory_stream_pos":
  235. # We need to make sure there is a single row, `(X, null), as that is
  236. # what synapse expects to be there.
  237. yield self.postgres_store._simple_insert(
  238. table=table,
  239. values={"stream_id": None},
  240. )
  241. self.progress.update(table, table_size) # Mark table as done
  242. return
  243. forward_select = (
  244. "SELECT rowid, * FROM %s WHERE rowid >= ? ORDER BY rowid LIMIT ?"
  245. % (table,)
  246. )
  247. backward_select = (
  248. "SELECT rowid, * FROM %s WHERE rowid <= ? ORDER BY rowid LIMIT ?"
  249. % (table,)
  250. )
  251. do_forward = [True]
  252. do_backward = [True]
  253. while True:
  254. def r(txn):
  255. forward_rows = []
  256. backward_rows = []
  257. if do_forward[0]:
  258. txn.execute(forward_select, (forward_chunk, self.batch_size,))
  259. forward_rows = txn.fetchall()
  260. if not forward_rows:
  261. do_forward[0] = False
  262. if do_backward[0]:
  263. txn.execute(backward_select, (backward_chunk, self.batch_size,))
  264. backward_rows = txn.fetchall()
  265. if not backward_rows:
  266. do_backward[0] = False
  267. if forward_rows or backward_rows:
  268. headers = [column[0] for column in txn.description]
  269. else:
  270. headers = None
  271. return headers, forward_rows, backward_rows
  272. headers, frows, brows = yield self.sqlite_store.runInteraction(
  273. "select", r
  274. )
  275. if frows or brows:
  276. if frows:
  277. forward_chunk = max(row[0] for row in frows) + 1
  278. if brows:
  279. backward_chunk = min(row[0] for row in brows) - 1
  280. rows = frows + brows
  281. rows = self._convert_rows(table, headers, rows)
  282. def insert(txn):
  283. self.postgres_store.insert_many_txn(
  284. txn, table, headers[1:], rows
  285. )
  286. self.postgres_store._simple_update_one_txn(
  287. txn,
  288. table="port_from_sqlite3",
  289. keyvalues={"table_name": table},
  290. updatevalues={
  291. "forward_rowid": forward_chunk,
  292. "backward_rowid": backward_chunk,
  293. },
  294. )
  295. yield self.postgres_store.execute(insert)
  296. postgres_size += len(rows)
  297. self.progress.update(table, postgres_size)
  298. else:
  299. return
  300. @defer.inlineCallbacks
  301. def handle_search_table(self, postgres_size, table_size, forward_chunk,
  302. backward_chunk):
  303. select = (
  304. "SELECT es.rowid, es.*, e.origin_server_ts, e.stream_ordering"
  305. " FROM event_search as es"
  306. " INNER JOIN events AS e USING (event_id, room_id)"
  307. " WHERE es.rowid >= ?"
  308. " ORDER BY es.rowid LIMIT ?"
  309. )
  310. while True:
  311. def r(txn):
  312. txn.execute(select, (forward_chunk, self.batch_size,))
  313. rows = txn.fetchall()
  314. headers = [column[0] for column in txn.description]
  315. return headers, rows
  316. headers, rows = yield self.sqlite_store.runInteraction("select", r)
  317. if rows:
  318. forward_chunk = rows[-1][0] + 1
  319. # We have to treat event_search differently since it has a
  320. # different structure in the two different databases.
  321. def insert(txn):
  322. sql = (
  323. "INSERT INTO event_search (event_id, room_id, key,"
  324. " sender, vector, origin_server_ts, stream_ordering)"
  325. " VALUES (?,?,?,?,to_tsvector('english', ?),?,?)"
  326. )
  327. rows_dict = []
  328. for row in rows:
  329. d = dict(zip(headers, row))
  330. if "\0" in d['value']:
  331. logger.warn('dropping search row %s', d)
  332. else:
  333. rows_dict.append(d)
  334. txn.executemany(sql, [
  335. (
  336. row["event_id"],
  337. row["room_id"],
  338. row["key"],
  339. row["sender"],
  340. row["value"],
  341. row["origin_server_ts"],
  342. row["stream_ordering"],
  343. )
  344. for row in rows_dict
  345. ])
  346. self.postgres_store._simple_update_one_txn(
  347. txn,
  348. table="port_from_sqlite3",
  349. keyvalues={"table_name": "event_search"},
  350. updatevalues={
  351. "forward_rowid": forward_chunk,
  352. "backward_rowid": backward_chunk,
  353. },
  354. )
  355. yield self.postgres_store.execute(insert)
  356. postgres_size += len(rows)
  357. self.progress.update("event_search", postgres_size)
  358. else:
  359. return
  360. def setup_db(self, db_config, database_engine):
  361. db_conn = database_engine.module.connect(
  362. **{
  363. k: v for k, v in db_config.get("args", {}).items()
  364. if not k.startswith("cp_")
  365. }
  366. )
  367. prepare_database(db_conn, database_engine, config=None)
  368. db_conn.commit()
  369. @defer.inlineCallbacks
  370. def run(self):
  371. try:
  372. sqlite_db_pool = adbapi.ConnectionPool(
  373. self.sqlite_config["name"],
  374. **self.sqlite_config["args"]
  375. )
  376. postgres_db_pool = adbapi.ConnectionPool(
  377. self.postgres_config["name"],
  378. **self.postgres_config["args"]
  379. )
  380. sqlite_engine = create_engine(sqlite_config)
  381. postgres_engine = create_engine(postgres_config)
  382. self.sqlite_store = Store(sqlite_db_pool, sqlite_engine)
  383. self.postgres_store = Store(postgres_db_pool, postgres_engine)
  384. yield self.postgres_store.execute(
  385. postgres_engine.check_database
  386. )
  387. # Step 1. Set up databases.
  388. self.progress.set_state("Preparing SQLite3")
  389. self.setup_db(sqlite_config, sqlite_engine)
  390. self.progress.set_state("Preparing PostgreSQL")
  391. self.setup_db(postgres_config, postgres_engine)
  392. # Step 2. Get tables.
  393. self.progress.set_state("Fetching tables")
  394. sqlite_tables = yield self.sqlite_store._simple_select_onecol(
  395. table="sqlite_master",
  396. keyvalues={
  397. "type": "table",
  398. },
  399. retcol="name",
  400. )
  401. postgres_tables = yield self.postgres_store._simple_select_onecol(
  402. table="information_schema.tables",
  403. keyvalues={},
  404. retcol="distinct table_name",
  405. )
  406. tables = set(sqlite_tables) & set(postgres_tables)
  407. self.progress.set_state("Creating tables")
  408. logger.info("Found %d tables", len(tables))
  409. def create_port_table(txn):
  410. txn.execute(
  411. "CREATE TABLE port_from_sqlite3 ("
  412. " table_name varchar(100) NOT NULL UNIQUE,"
  413. " forward_rowid bigint NOT NULL,"
  414. " backward_rowid bigint NOT NULL"
  415. ")"
  416. )
  417. # The old port script created a table with just a "rowid" column.
  418. # We want people to be able to rerun this script from an old port
  419. # so that they can pick up any missing events that were not
  420. # ported across.
  421. def alter_table(txn):
  422. txn.execute(
  423. "ALTER TABLE IF EXISTS port_from_sqlite3"
  424. " RENAME rowid TO forward_rowid"
  425. )
  426. txn.execute(
  427. "ALTER TABLE IF EXISTS port_from_sqlite3"
  428. " ADD backward_rowid bigint NOT NULL DEFAULT 0"
  429. )
  430. try:
  431. yield self.postgres_store.runInteraction(
  432. "alter_table", alter_table
  433. )
  434. except Exception as e:
  435. logger.info("Failed to create port table: %s", e)
  436. try:
  437. yield self.postgres_store.runInteraction(
  438. "create_port_table", create_port_table
  439. )
  440. except Exception as e:
  441. logger.info("Failed to create port table: %s", e)
  442. self.progress.set_state("Setting up")
  443. # Set up tables.
  444. setup_res = yield defer.gatherResults(
  445. [
  446. self.setup_table(table)
  447. for table in tables
  448. if table not in ["schema_version", "applied_schema_deltas"]
  449. and not table.startswith("sqlite_")
  450. ],
  451. consumeErrors=True,
  452. )
  453. # Process tables.
  454. yield defer.gatherResults(
  455. [
  456. self.handle_table(*res)
  457. for res in setup_res
  458. ],
  459. consumeErrors=True,
  460. )
  461. self.progress.done()
  462. except:
  463. global end_error_exec_info
  464. end_error_exec_info = sys.exc_info()
  465. logger.exception("")
  466. finally:
  467. reactor.stop()
  468. def _convert_rows(self, table, headers, rows):
  469. bool_col_names = BOOLEAN_COLUMNS.get(table, [])
  470. bool_cols = [
  471. i for i, h in enumerate(headers) if h in bool_col_names
  472. ]
  473. class BadValueException(Exception):
  474. pass
  475. def conv(j, col):
  476. if j in bool_cols:
  477. return bool(col)
  478. elif isinstance(col, basestring) and "\0" in col:
  479. logger.warn("DROPPING ROW: NUL value in table %s col %s: %r", table, headers[j], col)
  480. raise BadValueException();
  481. return col
  482. outrows = []
  483. for i, row in enumerate(rows):
  484. try:
  485. outrows.append(tuple(
  486. conv(j, col)
  487. for j, col in enumerate(row)
  488. if j > 0
  489. ))
  490. except BadValueException:
  491. pass
  492. return outrows
  493. @defer.inlineCallbacks
  494. def _setup_sent_transactions(self):
  495. # Only save things from the last day
  496. yesterday = int(time.time() * 1000) - 86400000
  497. # And save the max transaction id from each destination
  498. select = (
  499. "SELECT rowid, * FROM sent_transactions WHERE rowid IN ("
  500. "SELECT max(rowid) FROM sent_transactions"
  501. " GROUP BY destination"
  502. ")"
  503. )
  504. def r(txn):
  505. txn.execute(select)
  506. rows = txn.fetchall()
  507. headers = [column[0] for column in txn.description]
  508. ts_ind = headers.index('ts')
  509. return headers, [r for r in rows if r[ts_ind] < yesterday]
  510. headers, rows = yield self.sqlite_store.runInteraction(
  511. "select", r,
  512. )
  513. rows = self._convert_rows("sent_transactions", headers, rows)
  514. inserted_rows = len(rows)
  515. if inserted_rows:
  516. max_inserted_rowid = max(r[0] for r in rows)
  517. def insert(txn):
  518. self.postgres_store.insert_many_txn(
  519. txn, "sent_transactions", headers[1:], rows
  520. )
  521. yield self.postgres_store.execute(insert)
  522. else:
  523. max_inserted_rowid = 0
  524. def get_start_id(txn):
  525. txn.execute(
  526. "SELECT rowid FROM sent_transactions WHERE ts >= ?"
  527. " ORDER BY rowid ASC LIMIT 1",
  528. (yesterday,)
  529. )
  530. rows = txn.fetchall()
  531. if rows:
  532. return rows[0][0]
  533. else:
  534. return 1
  535. next_chunk = yield self.sqlite_store.execute(get_start_id)
  536. next_chunk = max(max_inserted_rowid + 1, next_chunk)
  537. yield self.postgres_store._simple_insert(
  538. table="port_from_sqlite3",
  539. values={
  540. "table_name": "sent_transactions",
  541. "forward_rowid": next_chunk,
  542. "backward_rowid": 0,
  543. }
  544. )
  545. def get_sent_table_size(txn):
  546. txn.execute(
  547. "SELECT count(*) FROM sent_transactions"
  548. " WHERE ts >= ?",
  549. (yesterday,)
  550. )
  551. size, = txn.fetchone()
  552. return int(size)
  553. remaining_count = yield self.sqlite_store.execute(
  554. get_sent_table_size
  555. )
  556. total_count = remaining_count + inserted_rows
  557. defer.returnValue((next_chunk, inserted_rows, total_count))
  558. @defer.inlineCallbacks
  559. def _get_remaining_count_to_port(self, table, forward_chunk, backward_chunk):
  560. frows = yield self.sqlite_store.execute_sql(
  561. "SELECT count(*) FROM %s WHERE rowid >= ?" % (table,),
  562. forward_chunk,
  563. )
  564. brows = yield self.sqlite_store.execute_sql(
  565. "SELECT count(*) FROM %s WHERE rowid <= ?" % (table,),
  566. backward_chunk,
  567. )
  568. defer.returnValue(frows[0][0] + brows[0][0])
  569. @defer.inlineCallbacks
  570. def _get_already_ported_count(self, table):
  571. rows = yield self.postgres_store.execute_sql(
  572. "SELECT count(*) FROM %s" % (table,),
  573. )
  574. defer.returnValue(rows[0][0])
  575. @defer.inlineCallbacks
  576. def _get_total_count_to_port(self, table, forward_chunk, backward_chunk):
  577. remaining, done = yield defer.gatherResults(
  578. [
  579. self._get_remaining_count_to_port(table, forward_chunk, backward_chunk),
  580. self._get_already_ported_count(table),
  581. ],
  582. consumeErrors=True,
  583. )
  584. remaining = int(remaining) if remaining else 0
  585. done = int(done) if done else 0
  586. defer.returnValue((done, remaining + done))
  587. ##############################################
  588. ###### The following is simply UI stuff ######
  589. ##############################################
  590. class Progress(object):
  591. """Used to report progress of the port
  592. """
  593. def __init__(self):
  594. self.tables = {}
  595. self.start_time = int(time.time())
  596. def add_table(self, table, cur, size):
  597. self.tables[table] = {
  598. "start": cur,
  599. "num_done": cur,
  600. "total": size,
  601. "perc": int(cur * 100 / size),
  602. }
  603. def update(self, table, num_done):
  604. data = self.tables[table]
  605. data["num_done"] = num_done
  606. data["perc"] = int(num_done * 100 / data["total"])
  607. def done(self):
  608. pass
  609. class CursesProgress(Progress):
  610. """Reports progress to a curses window
  611. """
  612. def __init__(self, stdscr):
  613. self.stdscr = stdscr
  614. curses.use_default_colors()
  615. curses.curs_set(0)
  616. curses.init_pair(1, curses.COLOR_RED, -1)
  617. curses.init_pair(2, curses.COLOR_GREEN, -1)
  618. self.last_update = 0
  619. self.finished = False
  620. self.total_processed = 0
  621. self.total_remaining = 0
  622. super(CursesProgress, self).__init__()
  623. def update(self, table, num_done):
  624. super(CursesProgress, self).update(table, num_done)
  625. self.total_processed = 0
  626. self.total_remaining = 0
  627. for table, data in self.tables.items():
  628. self.total_processed += data["num_done"] - data["start"]
  629. self.total_remaining += data["total"] - data["num_done"]
  630. self.render()
  631. def render(self, force=False):
  632. now = time.time()
  633. if not force and now - self.last_update < 0.2:
  634. # reactor.callLater(1, self.render)
  635. return
  636. self.stdscr.clear()
  637. rows, cols = self.stdscr.getmaxyx()
  638. duration = int(now) - int(self.start_time)
  639. minutes, seconds = divmod(duration, 60)
  640. duration_str = '%02dm %02ds' % (minutes, seconds,)
  641. if self.finished:
  642. status = "Time spent: %s (Done!)" % (duration_str,)
  643. else:
  644. if self.total_processed > 0:
  645. left = float(self.total_remaining) / self.total_processed
  646. est_remaining = (int(now) - self.start_time) * left
  647. est_remaining_str = '%02dm %02ds remaining' % divmod(est_remaining, 60)
  648. else:
  649. est_remaining_str = "Unknown"
  650. status = (
  651. "Time spent: %s (est. remaining: %s)"
  652. % (duration_str, est_remaining_str,)
  653. )
  654. self.stdscr.addstr(
  655. 0, 0,
  656. status,
  657. curses.A_BOLD,
  658. )
  659. max_len = max([len(t) for t in self.tables.keys()])
  660. left_margin = 5
  661. middle_space = 1
  662. items = self.tables.items()
  663. items.sort(
  664. key=lambda i: (i[1]["perc"], i[0]),
  665. )
  666. for i, (table, data) in enumerate(items):
  667. if i + 2 >= rows:
  668. break
  669. perc = data["perc"]
  670. color = curses.color_pair(2) if perc == 100 else curses.color_pair(1)
  671. self.stdscr.addstr(
  672. i + 2, left_margin + max_len - len(table),
  673. table,
  674. curses.A_BOLD | color,
  675. )
  676. size = 20
  677. progress = "[%s%s]" % (
  678. "#" * int(perc * size / 100),
  679. " " * (size - int(perc * size / 100)),
  680. )
  681. self.stdscr.addstr(
  682. i + 2, left_margin + max_len + middle_space,
  683. "%s %3d%% (%d/%d)" % (progress, perc, data["num_done"], data["total"]),
  684. )
  685. if self.finished:
  686. self.stdscr.addstr(
  687. rows - 1, 0,
  688. "Press any key to exit...",
  689. )
  690. self.stdscr.refresh()
  691. self.last_update = time.time()
  692. def done(self):
  693. self.finished = True
  694. self.render(True)
  695. self.stdscr.getch()
  696. def set_state(self, state):
  697. self.stdscr.clear()
  698. self.stdscr.addstr(
  699. 0, 0,
  700. state + "...",
  701. curses.A_BOLD,
  702. )
  703. self.stdscr.refresh()
  704. class TerminalProgress(Progress):
  705. """Just prints progress to the terminal
  706. """
  707. def update(self, table, num_done):
  708. super(TerminalProgress, self).update(table, num_done)
  709. data = self.tables[table]
  710. print "%s: %d%% (%d/%d)" % (
  711. table, data["perc"],
  712. data["num_done"], data["total"],
  713. )
  714. def set_state(self, state):
  715. print state + "..."
  716. ##############################################
  717. ##############################################
  718. if __name__ == "__main__":
  719. parser = argparse.ArgumentParser(
  720. description="A script to port an existing synapse SQLite database to"
  721. " a new PostgreSQL database."
  722. )
  723. parser.add_argument("-v", action='store_true')
  724. parser.add_argument(
  725. "--sqlite-database", required=True,
  726. help="The snapshot of the SQLite database file. This must not be"
  727. " currently used by a running synapse server"
  728. )
  729. parser.add_argument(
  730. "--postgres-config", type=argparse.FileType('r'), required=True,
  731. help="The database config file for the PostgreSQL database"
  732. )
  733. parser.add_argument(
  734. "--curses", action='store_true',
  735. help="display a curses based progress UI"
  736. )
  737. parser.add_argument(
  738. "--batch-size", type=int, default=1000,
  739. help="The number of rows to select from the SQLite table each"
  740. " iteration [default=1000]",
  741. )
  742. args = parser.parse_args()
  743. logging_config = {
  744. "level": logging.DEBUG if args.v else logging.INFO,
  745. "format": "%(asctime)s - %(name)s - %(lineno)d - %(levelname)s - %(message)s"
  746. }
  747. if args.curses:
  748. logging_config["filename"] = "port-synapse.log"
  749. logging.basicConfig(**logging_config)
  750. sqlite_config = {
  751. "name": "sqlite3",
  752. "args": {
  753. "database": args.sqlite_database,
  754. "cp_min": 1,
  755. "cp_max": 1,
  756. "check_same_thread": False,
  757. },
  758. }
  759. postgres_config = yaml.safe_load(args.postgres_config)
  760. if "database" in postgres_config:
  761. postgres_config = postgres_config["database"]
  762. if "name" not in postgres_config:
  763. sys.stderr.write("Malformed database config: no 'name'")
  764. sys.exit(2)
  765. if postgres_config["name"] != "psycopg2":
  766. sys.stderr.write("Database must use 'psycopg2' connector.")
  767. sys.exit(3)
  768. def start(stdscr=None):
  769. if stdscr:
  770. progress = CursesProgress(stdscr)
  771. else:
  772. progress = TerminalProgress()
  773. porter = Porter(
  774. sqlite_config=sqlite_config,
  775. postgres_config=postgres_config,
  776. progress=progress,
  777. batch_size=args.batch_size,
  778. )
  779. reactor.callWhenRunning(porter.run)
  780. reactor.run()
  781. if args.curses:
  782. curses.wrapper(start)
  783. else:
  784. start()
  785. if end_error_exec_info:
  786. exc_type, exc_value, exc_traceback = end_error_exec_info
  787. traceback.print_exception(exc_type, exc_value, exc_traceback)