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 ..db import status
  10 from ..db.connection import connect
  11 from ..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 def _osm2pgsql_options_from_args(args, default_cache, default_threads):
  21     """ Set up the standard osm2pgsql from the command line arguments.
  22     """
  23     return dict(osm2pgsql=args.osm2pgsql_path,
  24                 osm2pgsql_cache=args.osm2pgsql_cache or default_cache,
  25                 osm2pgsql_style=args.config.get_import_style_file(),
  26                 threads=args.threads or default_threads,
  27                 dsn=args.config.get_libpq_dsn(),
  28                 flatnode_file=args.config.FLATNODE_FILE)
  29
  30
  31 class UpdateReplication:
  32     """\
  33     Update the database using an online replication service.
  34     """
  35
  36     @staticmethod
  37     def add_args(parser):
  38         group = parser.add_argument_group('Arguments for initialisation')
  39         group.add_argument('--init', action='store_true',
  40                            help='Initialise the update process')
  41         group.add_argument('--no-update-functions', dest='update_functions',
  42                            action='store_false',
  43                            help="""Do not update the trigger function to
  44                                    support differential updates.""")
  45         group = parser.add_argument_group('Arguments for updates')
  46         group.add_argument('--check-for-updates', action='store_true',
  47                            help='Check if new updates are available and exit')
  48         group.add_argument('--once', action='store_true',
  49                            help="""Download and apply updates only once. When
  50                                    not set, updates are continuously applied""")
  51         group.add_argument('--no-index', action='store_false', dest='do_index',
  52                            help="""Do not index the new data. Only applicable
  53                                    together with --once""")
  54         group.add_argument('--osm2pgsql-cache', metavar='SIZE', type=int,
  55                            help='Size of cache to be used by osm2pgsql (in MB)')
  56         group = parser.add_argument_group('Download parameters')
  57         group.add_argument('--socket-timeout', dest='socket_timeout', type=int, default=60,
  58                            help='Set timeout for file downloads.')
  59
  60     @staticmethod
  61     def _init_replication(args):
  62         from ..tools import replication, refresh
  63
  64         LOG.warning("Initialising replication updates")
  65         conn = connect(args.config.get_libpq_dsn())
  66         replication.init_replication(conn, base_url=args.config.REPLICATION_URL)
  67         if args.update_functions:
  68             LOG.warning("Create functions")
  69             refresh.create_functions(conn, args.config, args.data_dir,
  70                                      True, False)
  71         conn.close()
  72         return 0
  73
  74
  75     @staticmethod
  76     def _check_for_updates(args):
  77         from ..tools import replication
  78
  79         conn = connect(args.config.get_libpq_dsn())
  80         ret = replication.check_for_updates(conn, base_url=args.config.REPLICATION_URL)
  81         conn.close()
  82         return ret
  83
  84     @staticmethod
  85     def _report_update(batchdate, start_import, start_index):
  86         def round_time(delta):
  87             return dt.timedelta(seconds=int(delta.total_seconds()))
  88
  89         end = dt.datetime.now(dt.timezone.utc)
  90         LOG.warning("Update completed. Import: %s. %sTotal: %s. Remaining backlog: %s.",
  91                     round_time((start_index or end) - start_import),
  92                     "Indexing: {} ".format(round_time(end - start_index))
  93                     if start_index else '',
  94                     round_time(end - start_import),
  95                     round_time(end - batchdate))
  96
  97     @staticmethod
  98     def _update(args):
  99         from ..tools import replication
 100         from ..indexer.indexer import Indexer
 101
 102         params = _osm2pgsql_options_from_args(args, 2000, 1)
 103         params.update(base_url=args.config.REPLICATION_URL,
 104                       update_interval=args.config.get_int('REPLICATION_UPDATE_INTERVAL'),
 105                       import_file=args.project_dir / 'osmosischange.osc',
 106                       max_diff_size=args.config.get_int('REPLICATION_MAX_DIFF'),
 107                       indexed_only=not args.once)
 108
 109         # Sanity check to not overwhelm the Geofabrik servers.
 110         if 'download.geofabrik.de'in params['base_url']\
 111            and params['update_interval'] < 86400:
 112             LOG.fatal("Update interval too low for download.geofabrik.de.\n"
 113                       "Please check install documentation "
 114                       "(https://nominatim.org/release-docs/latest/admin/Import-and-Update#"
 115                       "setting-up-the-update-process).")
 116             raise UsageError("Invalid replication update interval setting.")
 117
 118         if not args.once:
 119             if not args.do_index:
 120                 LOG.fatal("Indexing cannot be disabled when running updates continuously.")
 121                 raise UsageError("Bad argument '--no-index'.")
 122             recheck_interval = args.config.get_int('REPLICATION_RECHECK_INTERVAL')
 123
 124         while True:
 125             conn = connect(args.config.get_libpq_dsn())
 126             start = dt.datetime.now(dt.timezone.utc)
 127             state = replication.update(conn, params)
 128             if state is not replication.UpdateState.NO_CHANGES:
 129                 status.log_status(conn, start, 'import')
 130             batchdate, _, _ = status.get_status(conn)
 131             conn.close()
 132
 133             if state is not replication.UpdateState.NO_CHANGES and args.do_index:
 134                 index_start = dt.datetime.now(dt.timezone.utc)
 135                 indexer = Indexer(args.config.get_libpq_dsn(),
 136                                   args.threads or 1)
 137                 indexer.index_boundaries(0, 30)
 138                 indexer.index_by_rank(0, 30)
 139
 140                 conn = connect(args.config.get_libpq_dsn())
 141                 status.set_indexed(conn, True)
 142                 status.log_status(conn, index_start, 'index')
 143                 conn.close()
 144             else:
 145                 index_start = None
 146
 147             if LOG.isEnabledFor(logging.WARNING):
 148                 UpdateReplication._report_update(batchdate, start, index_start)
 149
 150             if args.once:
 151                 break
 152
 153             if state is replication.UpdateState.NO_CHANGES:
 154                 LOG.warning("No new changes. Sleeping for %d sec.", recheck_interval)
 155                 time.sleep(recheck_interval)
 156
 157         return state.value
 158
 159
 160     @staticmethod
 161     def run(args):
 162         socket.setdefaulttimeout(args.socket_timeout)
 163
 164         if args.init:
 165             return UpdateReplication._init_replication(args)
 166
 167         if args.check_for_updates:
 168             return UpdateReplication._check_for_updates(args)
 169
 170         return UpdateReplication._update(args)