Merge remote-tracking branch 'upstream/master'

[nominatim.git] / nominatim / db / async_connection.py

diff --git a/nominatim/db/async_connection.py b/nominatim/db/async_connection.py
index 85b844312ba271a638be8ef084dbe34a9274889d..285463a50ffd9d7eee8efd1fa931d0bdcc4254f7 100644 (file)
--- a/nominatim/db/async_connection.py
+++ b/nominatim/db/async_connection.py
@@ -1,55 +1,95 @@
 # SPDX-License-Identifier: GPL-2.0-only
 #
 # 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.
 """
 import logging
 # For a full list of authors see the git log.
 """ Database helper functions for the indexer.
 """
 import logging

+import select
+import time
+

 import psycopg2
 from psycopg2.extras import wait_select
 
 import psycopg2
 from psycopg2.extras import wait_select
 

+# psycopg2 emits different exceptions pre and post 2.8. Detect if the new error
+# module is available and adapt the error handling accordingly.
+try:
+    import psycopg2.errors # pylint: disable=no-name-in-module,import-error
+    __has_psycopg2_errors__ = True
+except ImportError:
+    __has_psycopg2_errors__ = False
+

 LOG = logging.getLogger()
 
 LOG = logging.getLogger()
 

-def make_connection(options, asynchronous=False):
-    """ Create a psycopg2 connection from the given options.
+class DeadlockHandler:
+    """ Context manager that catches deadlock exceptions and calls
+        the given handler function. All other exceptions are passed on
+        normally.

     """
     """

-    params = {'dbname' : options.dbname,
-              'user' : options.user,
-              'password' : options.password,
-              'host' : options.host,
-              'port' : options.port,
-              'async' : asynchronous}

 
 

-    return psycopg2.connect(**params)
+    def __init__(self, handler, ignore_sql_errors=False):
+        self.handler = handler
+        self.ignore_sql_errors = ignore_sql_errors
+
+    def __enter__(self):
+        return self
+
+    def __exit__(self, exc_type, exc_value, traceback):
+        if __has_psycopg2_errors__:
+            if exc_type == psycopg2.errors.DeadlockDetected: # pylint: disable=E1101
+                self.handler()
+                return True
+        elif exc_type == psycopg2.extensions.TransactionRollbackError \
+             and exc_value.pgcode == '40P01':
+            self.handler()
+            return True
+
+        if self.ignore_sql_errors and isinstance(exc_value, psycopg2.Error):
+            LOG.info("SQL error ignored: %s", exc_value)
+            return True
+
+        return False
+

 
 class DBConnection:
     """ A single non-blocking database connection.
     """
 
 
 class DBConnection:
     """ A single non-blocking database connection.
     """
 

-    def __init__(self, options):
+    def __init__(self, dsn, cursor_factory=None, ignore_sql_errors=False):

         self.current_query = None
         self.current_params = None
         self.current_query = None
         self.current_params = None

-        self.options = options
+        self.dsn = dsn
+        self.ignore_sql_errors = ignore_sql_errors

 
         self.conn = None
         self.cursor = None
 
         self.conn = None
         self.cursor = None

-        self.connect()
+        self.connect(cursor_factory=cursor_factory)

 
 

-    def connect(self):
+    def close(self):
+        """ Close all open connections. Does not wait for pending requests.
+        """
+        if self.conn is not None:
+            self.cursor.close()
+            self.conn.close()
+
+        self.conn = None
+
+    def connect(self, cursor_factory=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.
             The caller must ensure that no query is pending before reconnecting.
         """
         """ (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.
             The caller must ensure that no query is pending before reconnecting.
         """

-        if self.conn is not None:
-            self.cursor.close()
-            self.conn.close()
+        self.close()

 
 

-        self.conn = make_connection(self.options, asynchronous=True)
+        # 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.wait()
 
         self.wait()
 

-        self.cursor = self.conn.cursor()
+        self.cursor = self.conn.cursor(cursor_factory=cursor_factory)

         # 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
         # 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


@@ -60,23 +100,18 @@ class DBConnection:
                    WHERE name = 'max_parallel_workers_per_gather';""")
         self.wait()
 
                    WHERE name = 'max_parallel_workers_per_gather';""")
         self.wait()
 

+    def _deadlock_handler(self):
+        LOG.info("Deadlock detected (params = %s), retry.", str(self.current_params))
+        self.cursor.execute(self.current_query, self.current_params)
+

     def wait(self):
         """ Block until any pending operation is done.
         """
         while True:
     def wait(self):
         """ Block until any pending operation is done.
         """
         while True:

-            try:
+            with DeadlockHandler(self._deadlock_handler, self.ignore_sql_errors):

                 wait_select(self.conn)
                 self.current_query = None
                 return
                 wait_select(self.conn)
                 self.current_query = None
                 return

-            except psycopg2.extensions.TransactionRollbackError as error:
-                if error.pgcode == '40P01':
-                    LOG.info("Deadlock detected (params = %s), retry.",
-                             str(self.current_params))
-                    self.cursor.execute(self.current_query, self.current_params)
-                else:
-                    raise
-            except psycopg2.errors.DeadlockDetected: # pylint: disable=E1101
-                self.cursor.execute(self.current_query, self.current_params)

 
     def perform(self, sql, args=None):
         """ Send SQL query to the server. Returns immediately without
 
     def perform(self, sql, args=None):
         """ Send SQL query to the server. Returns immediately without


@@ -100,17 +135,82 @@ class DBConnection:
         if self.current_query is None:
             return True
 
         if self.current_query is None:
             return True
 

-        try:
+        with DeadlockHandler(self._deadlock_handler, self.ignore_sql_errors):

             if self.conn.poll() == psycopg2.extensions.POLL_OK:
                 self.current_query = None
                 return True
             if self.conn.poll() == psycopg2.extensions.POLL_OK:
                 self.current_query = None
                 return True

-        except psycopg2.extensions.TransactionRollbackError as error:
-            if error.pgcode == '40P01':
-                LOG.info("Deadlock detected (params = %s), retry.", str(self.current_params))
-                self.cursor.execute(self.current_query, self.current_params)
-            else:
-                raise
-        except psycopg2.errors.DeadlockDetected: # pylint: disable=E1101
-            self.cursor.execute(self.current_query, self.current_params)

 
         return False
 
         return False

+
+
+class WorkerPool:
+    """ A pool of asynchronous database connections.
+
+        The pool may be used as a context manager.
+    """
+    REOPEN_CONNECTIONS_AFTER = 100000
+
+    def __init__(self, dsn, pool_size, ignore_sql_errors=False):
+        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
+
+
+    def finish_all(self):
+        """ Wait for all connection to finish.
+        """
+        for thread in self.threads:
+            while not thread.is_done():
+                thread.wait()
+
+        self.free_workers = self._yield_free_worker()
+
+    def close(self):
+        """ Close all connections and clear the pool.
+        """
+        for thread in self.threads:
+            thread.close()
+        self.threads = []
+        self.free_workers = None
+
+
+    def next_free_worker(self):
+        """ Get the next free connection.
+        """
+        return next(self.free_workers)
+
+
+    def _yield_free_worker(self):
+        ready = self.threads
+        command_stat = 0
+        while True:
+            for thread in ready:
+                if thread.is_done():
+                    command_stat += 1
+                    yield thread
+
+            if command_stat > self.REOPEN_CONNECTIONS_AFTER:
+                self._reconnect_threads()
+                ready = self.threads
+                command_stat = 0
+            else:
+                tstart = time.time()
+                _, ready, _ = select.select([], self.threads, [])
+                self.wait_time += time.time() - tstart
+
+
+    def _reconnect_threads(self):
+        for thread in self.threads:
+            while not thread.is_done():
+                thread.wait()
+            thread.connect()
+
+
+    def __enter__(self):
+        return self
+
+
+    def __exit__(self, exc_type, exc_value, traceback):
+        self.finish_all()
+        self.close()