nominatim/clicmd/replication.py

   1 """
   2 Implementation of the 'replication' sub-command.
   3 """
   4 import datetime as dt
   5 import logging
   6 import socket
   7 import time
   8
   9 from nominatim.db import status
  10 from nominatim.db.connection import connect
  11 from nominatim.errors import UsageError
  12
  13 LOG = logging.getLogger()
  14
  15 # Do not repeat documentation of subcommand classes.
  16 # pylint: disable=C0111
  17 # Using non-top-level imports to make pyosmium optional for replication only.
  18 # pylint: disable=E0012,C0415
  19
  20 class UpdateReplication:
  21     """\
  22     Update the database using an online replication service.
  23     """
  24
  25     @staticmethod
  26     def add_args(parser):
  27         group = parser.add_argument_group('Arguments for initialisation')
  28         group.add_argument('--init', action='store_true',
  29                            help='Initialise the update process')
  30         group.add_argument('--no-update-functions', dest='update_functions',
  31                            action='store_false',
  32                            help=("Do not update the trigger function to "
  33                                  "support differential updates."))
  34         group = parser.add_argument_group('Arguments for updates')
  35         group.add_argument('--check-for-updates', action='store_true',
  36                            help='Check if new updates are available and exit')
  37         group.add_argument('--once', action='store_true',
  38                            help=("Download and apply updates only once. When "
  39                                  "not set, updates are continuously applied"))
  40         group.add_argument('--no-index', action='store_false', dest='do_index',
  41                            help=("Do not index the new data. Only applicable "
  42                                  "together with --once"))
  43         group.add_argument('--osm2pgsql-cache', metavar='SIZE', type=int,
  44                            help='Size of cache to be used by osm2pgsql (in MB)')
  45         group = parser.add_argument_group('Download parameters')
  46         group.add_argument('--socket-timeout', dest='socket_timeout', type=int, default=60,
  47                            help='Set timeout for file downloads.')
  48
  49     @staticmethod
  50     def _init_replication(args):
  51         from ..tools import replication, refresh
  52
  53         LOG.warning("Initialising replication updates")
  54         with connect(args.config.get_libpq_dsn()) as conn:
  55             replication.init_replication(conn, base_url=args.config.REPLICATION_URL)
  56             if args.update_functions:
  57                 LOG.warning("Create functions")
  58                 refresh.create_functions(conn, args.config, True, False)
  59         return 0
  60
  61
  62     @staticmethod
  63     def _check_for_updates(args):
  64         from ..tools import replication
  65
  66         with connect(args.config.get_libpq_dsn()) as conn:
  67             return replication.check_for_updates(conn, base_url=args.config.REPLICATION_URL)
  68
  69     @staticmethod
  70     def _report_update(batchdate, start_import, start_index):
  71         def round_time(delta):
  72             return dt.timedelta(seconds=int(delta.total_seconds()))
  73
  74         end = dt.datetime.now(dt.timezone.utc)
  75         LOG.warning("Update completed. Import: %s. %sTotal: %s. Remaining backlog: %s.",
  76                     round_time((start_index or end) - start_import),
  77                     "Indexing: {} ".format(round_time(end - start_index))
  78                     if start_index else '',
  79                     round_time(end - start_import),
  80                     round_time(end - batchdate))
  81
  82     @staticmethod
  83     def _update(args):
  84         from ..tools import replication
  85         from ..indexer.indexer import Indexer
  86         from ..tokenizer import factory as tokenizer_factory
  87
  88         params = args.osm2pgsql_options(default_cache=2000, default_threads=1)
  89         params.update(base_url=args.config.REPLICATION_URL,
  90                       update_interval=args.config.get_int('REPLICATION_UPDATE_INTERVAL'),
  91                       import_file=args.project_dir / 'osmosischange.osc',
  92                       max_diff_size=args.config.get_int('REPLICATION_MAX_DIFF'),
  93                       indexed_only=not args.once)
  94
  95         # Sanity check to not overwhelm the Geofabrik servers.
  96         if 'download.geofabrik.de' in params['base_url']\
  97            and params['update_interval'] < 86400:
  98             LOG.fatal("Update interval too low for download.geofabrik.de.\n"
  99                       "Please check install documentation "
 100                       "(https://nominatim.org/release-docs/latest/admin/Import-and-Update#"
 101                       "setting-up-the-update-process).")
 102             raise UsageError("Invalid replication update interval setting.")
 103
 104         if not args.once:
 105             if not args.do_index:
 106                 LOG.fatal("Indexing cannot be disabled when running updates continuously.")
 107                 raise UsageError("Bad argument '--no-index'.")
 108             recheck_interval = args.config.get_int('REPLICATION_RECHECK_INTERVAL')
 109
 110         tokenizer = tokenizer_factory.get_tokenizer_for_db(args.config)
 111
 112         while True:
 113             with connect(args.config.get_libpq_dsn()) as conn:
 114                 start = dt.datetime.now(dt.timezone.utc)
 115                 state = replication.update(conn, params)
 116                 if state is not replication.UpdateState.NO_CHANGES:
 117                     status.log_status(conn, start, 'import')
 118                 batchdate, _, _ = status.get_status(conn)
 119                 conn.commit()
 120
 121             if state is not replication.UpdateState.NO_CHANGES and args.do_index:
 122                 index_start = dt.datetime.now(dt.timezone.utc)
 123                 indexer = Indexer(args.config.get_libpq_dsn(), tokenizer,
 124                                   args.threads or 1)
 125                 indexer.index_boundaries(0, 30)
 126                 indexer.index_by_rank(0, 30)
 127
 128                 with connect(args.config.get_libpq_dsn()) as conn:
 129                     status.set_indexed(conn, True)
 130                     status.log_status(conn, index_start, 'index')
 131                     conn.commit()
 132             else:
 133                 index_start = None
 134
 135             if LOG.isEnabledFor(logging.WARNING):
 136                 UpdateReplication._report_update(batchdate, start, index_start)
 137
 138             if args.once:
 139                 break
 140
 141             if state is replication.UpdateState.NO_CHANGES:
 142                 LOG.warning("No new changes. Sleeping for %d sec.", recheck_interval)
 143                 time.sleep(recheck_interval)
 144
 145
 146     @staticmethod
 147     def run(args):
 148         socket.setdefaulttimeout(args.socket_timeout)
 149
 150         if args.init:
 151             return UpdateReplication._init_replication(args)
 152
 153         if args.check_for_updates:
 154             return UpdateReplication._check_for_updates(args)
 155
 156         UpdateReplication._update(args)
 157         return 0