Replication process failing

[tlaugier]

Hi, I have a Nominatim instance running with the nominatim-docker image. For about a week now I noticed that my data are not updating anymore.
When I try to start the replication process manually with docker exec -itu nominatim nominatim nominatim replication --project-dir=/nominatim --catch-up I get this error:

2026-06-01 06:32:29: Using project directory: /nominatim
2026-06-01 06:33:41  osm2pgsql version 1.11.0
2026-06-01 06:33:41  Database version: 16.11 (Ubuntu 16.11-0ubuntu0.24.04.1)
2026-06-01 06:33:41  PostGIS version: 3.4
2026-06-01 06:33:41  Loading properties from table '"public"."osm2pgsql_properties"'.
2026-06-01 06:33:41  Using flat node file '/nominatim/flatnode/flatnode.file' (same as on import).
2026-06-01 06:33:41  Using prefix 'planet_osm' (same as on import).
2026-06-01 06:33:41  Using style file '/usr/local/lib/python3.12/dist-packages/nominatim_db/resources/lib-lua/import-address.lua' (same as on import).
Processing: Node(11039k 113.8k/s) Way(3257k 93.06k/s) Relation(0 0.0/s))
2026-06-01 06:36:19  ERROR: DB copy thread failed: Database error: ERROR:  invalid memory alloc request size 1073741824

Traceback (most recent call last):
  File "/usr/local/bin/nominatim", line 5, in <module>
    exit(cli.nominatim())
         ^^^^^^^^^^^^^^^
  File "/usr/local/lib/python3.12/dist-packages/nominatim_db/cli.py", line 244, in nominatim
    return get_set_parser().run(cli_args=cli_args, environ=environ)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/usr/local/lib/python3.12/dist-packages/nominatim_db/cli.py", line 116, in run
    return args.command.run(args)
           ^^^^^^^^^^^^^^^^^^^^^^
  File "/usr/local/lib/python3.12/dist-packages/nominatim_db/clicmd/replication.py", line 188, in run
    asyncio.run(self._update(args))
  File "/usr/lib/python3.12/asyncio/runners.py", line 194, in run
    return runner.run(main)
           ^^^^^^^^^^^^^^^^
  File "/usr/lib/python3.12/asyncio/runners.py", line 118, in run
    return self._loop.run_until_complete(task)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/usr/lib/python3.12/asyncio/base_events.py", line 687, in run_until_complete
    return future.result()
           ^^^^^^^^^^^^^^^
  File "/usr/local/lib/python3.12/dist-packages/nominatim_db/clicmd/replication.py", line 145, in _update
    state = replication.update(dsn, params, socket_timeout=args.socket_timeout)
            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/usr/local/lib/python3.12/dist-packages/nominatim_db/tools/replication.py", line 138, in update
    run_osm2pgsql_updates(conn, options)
  File "/usr/local/lib/python3.12/dist-packages/nominatim_db/tools/replication.py", line 160, in run_osm2pgsql_updates
    run_osm2pgsql(options)
  File "/usr/local/lib/python3.12/dist-packages/nominatim_db/tools/exec_utils.py", line 68, in run_osm2pgsql
    subprocess.run(cmd, cwd=options.get('cwd', '.'),
  File "/usr/lib/python3.12/subprocess.py", line 571, in run
    raise CalledProcessError(retcode, process.args,
subprocess.CalledProcessError: Command '['/usr/bin/osm2pgsql', '--append', '--slim', '--log-progress', 'true', '--number-processes', '1', '--cache', '2000', '--style', '/usr/local/lib/python3.12/dist-packages/nominatim_db/resources/lib-lua/import-address.lua', '--output', 'flex', '--flat-nodes', '/nominatim/flatnode/flatnode.file', '/nominatim/osmosischange.osc']' returned non-zero exit status 2.

Searching the web lead me to increase the value for the shared_buffers setting of PostgreSQL to 32GB but this did not fix the issue.

Does anybody has encountered a similar issue? I'd really like to be able to restart the replication as soon as possible and not have to do a full rebuild if I fall too much behind.

[dudrv]

Hi, did you find a solution? I bumped into this also. I tried to run update with --once only in hope it will process less data, but still too much memory allocated.

[tlaugier]

Unfortunately no I haven't found a solution for this. I tried multiple times to restart an update using the various arguments but none succeeded. We since decided to move on to using Pelias so I gave up on updating this server.