X-Git-Url: https://git.openstreetmap.org./nominatim.git/blobdiff_plain/b4fec57b6d53f8e8a45c46ff11f13cbcbea1006a..c41f2fed2133668dc3179813261d39d3ff69cbdd:/nominatim/db/async_connection.py?ds=sidebyside diff --git a/nominatim/db/async_connection.py b/nominatim/db/async_connection.py index f06f3159..07767e48 100644 --- a/nominatim/db/async_connection.py +++ b/nominatim/db/async_connection.py @@ -1,10 +1,12 @@ # SPDX-License-Identifier: GPL-2.0-only # -# This file is part of Nominatim. -# Copyright (C) 2021 by the Nominatim developer community. +# This file is part of Nominatim. (https://nominatim.org) +# +# Copyright (C) 2022 by the Nominatim developer community. # For a full list of authors see the git log. -""" Database helper functions for the indexer. +""" Non-blocking database connections. """ +from typing import Callable, Any, Optional, Iterator, Sequence import logging import select import time @@ -20,6 +22,8 @@ try: except ImportError: __has_psycopg2_errors__ = False +from nominatim.typing import T_cursor, Query + LOG = logging.getLogger() class DeadlockHandler: @@ -28,14 +32,14 @@ class DeadlockHandler: normally. """ - def __init__(self, handler, ignore_sql_errors=False): + def __init__(self, handler: Callable[[], None], ignore_sql_errors: bool = False) -> None: self.handler = handler self.ignore_sql_errors = ignore_sql_errors - def __enter__(self): + def __enter__(self) -> 'DeadlockHandler': return self - def __exit__(self, exc_type, exc_value, traceback): + def __exit__(self, exc_type: Any, exc_value: Any, traceback: Any) -> bool: if __has_psycopg2_errors__: if exc_type == psycopg2.errors.DeadlockDetected: # pylint: disable=E1101 self.handler() @@ -56,26 +60,31 @@ class DBConnection: """ A single non-blocking database connection. """ - def __init__(self, dsn, cursor_factory=None, ignore_sql_errors=False): - self.current_query = None - self.current_params = None + def __init__(self, dsn: str, + cursor_factory: Optional[Callable[..., T_cursor]] = None, + ignore_sql_errors: bool = False) -> None: self.dsn = dsn + + self.current_query: Optional[Query] = None + self.current_params: Optional[Sequence[Any]] = None self.ignore_sql_errors = ignore_sql_errors - self.conn = None - self.cursor = None + self.conn: Optional['psycopg2._psycopg.connection'] = None + self.cursor: Optional['psycopg2._psycopg.cursor'] = None self.connect(cursor_factory=cursor_factory) - def close(self): + def close(self) -> None: """ Close all open connections. Does not wait for pending requests. """ if self.conn is not None: - self.cursor.close() + if self.cursor is not None: + self.cursor.close() + self.cursor = None self.conn.close() self.conn = None - def connect(self, cursor_factory=None): + def connect(self, cursor_factory: Optional[Callable[..., T_cursor]] = None) -> None: """ (Re)connect to the database. Creates an asynchronous connection with JIT and parallel processing disabled. If a connection was already open, it is closed and a new connection established. @@ -85,10 +94,14 @@ class DBConnection: # Use a dict to hand in the parameters because async is a reserved # word in Python3. - self.conn = psycopg2.connect(**{'dsn' : self.dsn, 'async' : True}) + self.conn = psycopg2.connect(**{'dsn': self.dsn, 'async': True}) # type: ignore + assert self.conn self.wait() - self.cursor = self.conn.cursor(cursor_factory=cursor_factory) + if cursor_factory is not None: + self.cursor = self.conn.cursor(cursor_factory=cursor_factory) + else: + self.cursor = self.conn.cursor() # Disable JIT and parallel workers as they are known to cause problems. # Update pg_settings instead of using SET because it does not yield # errors on older versions of Postgres where the settings are not @@ -99,11 +112,15 @@ class DBConnection: WHERE name = 'max_parallel_workers_per_gather';""") self.wait() - def _deadlock_handler(self): + def _deadlock_handler(self) -> None: LOG.info("Deadlock detected (params = %s), retry.", str(self.current_params)) + assert self.cursor is not None + assert self.current_query is not None + assert self.current_params is not None + self.cursor.execute(self.current_query, self.current_params) - def wait(self): + def wait(self) -> None: """ Block until any pending operation is done. """ while True: @@ -112,25 +129,29 @@ class DBConnection: self.current_query = None return - def perform(self, sql, args=None): + def perform(self, sql: Query, args: Optional[Sequence[Any]] = None) -> None: """ Send SQL query to the server. Returns immediately without blocking. """ + assert self.cursor is not None self.current_query = sql self.current_params = args self.cursor.execute(sql, args) - def fileno(self): + def fileno(self) -> int: """ File descriptor to wait for. (Makes this class select()able.) """ + assert self.conn is not None return self.conn.fileno() - def is_done(self): + def is_done(self) -> bool: """ Check if the connection is available for a new query. Also checks if the previous query has run into a deadlock. If so, then the previous query is repeated. """ + assert self.conn is not None + if self.current_query is None: return True @@ -149,14 +170,14 @@ class WorkerPool: """ REOPEN_CONNECTIONS_AFTER = 100000 - def __init__(self, dsn, pool_size, ignore_sql_errors=False): + def __init__(self, dsn: str, pool_size: int, ignore_sql_errors: bool = False) -> None: self.threads = [DBConnection(dsn, ignore_sql_errors=ignore_sql_errors) for _ in range(pool_size)] self.free_workers = self._yield_free_worker() - self.wait_time = 0 + self.wait_time = 0.0 - def finish_all(self): + def finish_all(self) -> None: """ Wait for all connection to finish. """ for thread in self.threads: @@ -165,22 +186,22 @@ class WorkerPool: self.free_workers = self._yield_free_worker() - def close(self): + def close(self) -> None: """ Close all connections and clear the pool. """ for thread in self.threads: thread.close() self.threads = [] - self.free_workers = None + self.free_workers = iter([]) - def next_free_worker(self): + def next_free_worker(self) -> DBConnection: """ Get the next free connection. """ return next(self.free_workers) - def _yield_free_worker(self): + def _yield_free_worker(self) -> Iterator[DBConnection]: ready = self.threads command_stat = 0 while True: @@ -199,17 +220,17 @@ class WorkerPool: self.wait_time += time.time() - tstart - def _reconnect_threads(self): + def _reconnect_threads(self) -> None: for thread in self.threads: while not thread.is_done(): thread.wait() thread.connect() - def __enter__(self): + def __enter__(self) -> 'WorkerPool': return self - def __exit__(self, exc_type, exc_value, traceback): + def __exit__(self, exc_type: Any, exc_value: Any, traceback: Any) -> None: self.finish_all() self.close()