Auto repack the database on the second sunday of the month (#69314)

tests/components/recorder/test_init.py

@@ -12,6 +12,7 @@ from sqlalchemy.exc import DatabaseError, OperationalError, SQLAlchemyError
 from homeassistant.components import recorder
 from homeassistant.components.recorder import (
     CONF_AUTO_PURGE,
+    CONF_AUTO_REPACK,
     CONF_DB_URL,
     CONFIG_SCHEMA,
     DOMAIN,
@@ -70,6 +71,7 @@ def _default_recorder(hass):
     return Recorder(
         hass,
         auto_purge=True,
+        auto_repack=True,
         keep_days=7,
         commit_interval=1,
         uri="sqlite://",
@@ -627,6 +629,7 @@ async def test_defaults_set(hass):
     assert recorder_config is not None
     # pylint: disable=unsubscriptable-object
     assert recorder_config["auto_purge"]
+    assert recorder_config["auto_repack"]
     assert recorder_config["purge_keep_days"] == 10
 
 
@@ -695,6 +698,120 @@ def test_auto_purge(hass_recorder):
     dt_util.set_default_time_zone(original_tz)
 
 
+@pytest.mark.parametrize("enable_nightly_purge", [True])
+def test_auto_purge_auto_repack_on_second_sunday(hass_recorder):
+    """Test periodic purge scheduling does a repack on the 2nd sunday."""
+    hass = hass_recorder()
+
+    original_tz = dt_util.DEFAULT_TIME_ZONE
+
+    tz = dt_util.get_time_zone("Europe/Copenhagen")
+    dt_util.set_default_time_zone(tz)
+
+    # Purging is scheduled to happen at 4:12am every day. Exercise this behavior by
+    # firing time changed events and advancing the clock around this time. Pick an
+    # arbitrary year in the future to avoid boundary conditions relative to the current
+    # date.
+    #
+    # The clock is started at 4:15am then advanced forward below
+    now = dt_util.utcnow()
+    test_time = datetime(now.year + 2, 1, 1, 4, 15, 0, tzinfo=tz)
+    run_tasks_at_time(hass, test_time)
+
+    with patch(
+        "homeassistant.components.recorder.is_second_sunday", return_value=True
+    ), patch(
+        "homeassistant.components.recorder.purge.purge_old_data", return_value=True
+    ) as purge_old_data, patch(
+        "homeassistant.components.recorder.perodic_db_cleanups"
+    ) as perodic_db_cleanups:
+        # Advance one day, and the purge task should run
+        test_time = test_time + timedelta(days=1)
+        run_tasks_at_time(hass, test_time)
+        assert len(purge_old_data.mock_calls) == 1
+        args, _ = purge_old_data.call_args_list[0]
+        assert args[2] is True  # repack
+        assert len(perodic_db_cleanups.mock_calls) == 1
+
+    dt_util.set_default_time_zone(original_tz)
+
+
+@pytest.mark.parametrize("enable_nightly_purge", [True])
+def test_auto_purge_auto_repack_disabled_on_second_sunday(hass_recorder):
+    """Test periodic purge scheduling does not auto repack on the 2nd sunday if disabled."""
+    hass = hass_recorder({CONF_AUTO_REPACK: False})
+
+    original_tz = dt_util.DEFAULT_TIME_ZONE
+
+    tz = dt_util.get_time_zone("Europe/Copenhagen")
+    dt_util.set_default_time_zone(tz)
+
+    # Purging is scheduled to happen at 4:12am every day. Exercise this behavior by
+    # firing time changed events and advancing the clock around this time. Pick an
+    # arbitrary year in the future to avoid boundary conditions relative to the current
+    # date.
+    #
+    # The clock is started at 4:15am then advanced forward below
+    now = dt_util.utcnow()
+    test_time = datetime(now.year + 2, 1, 1, 4, 15, 0, tzinfo=tz)
+    run_tasks_at_time(hass, test_time)
+
+    with patch(
+        "homeassistant.components.recorder.is_second_sunday", return_value=True
+    ), patch(
+        "homeassistant.components.recorder.purge.purge_old_data", return_value=True
+    ) as purge_old_data, patch(
+        "homeassistant.components.recorder.perodic_db_cleanups"
+    ) as perodic_db_cleanups:
+        # Advance one day, and the purge task should run
+        test_time = test_time + timedelta(days=1)
+        run_tasks_at_time(hass, test_time)
+        assert len(purge_old_data.mock_calls) == 1
+        args, _ = purge_old_data.call_args_list[0]
+        assert args[2] is False  # repack
+        assert len(perodic_db_cleanups.mock_calls) == 1
+
+    dt_util.set_default_time_zone(original_tz)
+
+
+@pytest.mark.parametrize("enable_nightly_purge", [True])
+def test_auto_purge_no_auto_repack_on_not_second_sunday(hass_recorder):
+    """Test periodic purge scheduling does not do a repack unless its the 2nd sunday."""
+    hass = hass_recorder()
+
+    original_tz = dt_util.DEFAULT_TIME_ZONE
+
+    tz = dt_util.get_time_zone("Europe/Copenhagen")
+    dt_util.set_default_time_zone(tz)
+
+    # Purging is scheduled to happen at 4:12am every day. Exercise this behavior by
+    # firing time changed events and advancing the clock around this time. Pick an
+    # arbitrary year in the future to avoid boundary conditions relative to the current
+    # date.
+    #
+    # The clock is started at 4:15am then advanced forward below
+    now = dt_util.utcnow()
+    test_time = datetime(now.year + 2, 1, 1, 4, 15, 0, tzinfo=tz)
+    run_tasks_at_time(hass, test_time)
+
+    with patch(
+        "homeassistant.components.recorder.is_second_sunday", return_value=False
+    ), patch(
+        "homeassistant.components.recorder.purge.purge_old_data", return_value=True
+    ) as purge_old_data, patch(
+        "homeassistant.components.recorder.perodic_db_cleanups"
+    ) as perodic_db_cleanups:
+        # Advance one day, and the purge task should run
+        test_time = test_time + timedelta(days=1)
+        run_tasks_at_time(hass, test_time)
+        assert len(purge_old_data.mock_calls) == 1
+        args, _ = purge_old_data.call_args_list[0]
+        assert args[2] is False  # repack
+        assert len(perodic_db_cleanups.mock_calls) == 1
+
+    dt_util.set_default_time_zone(original_tz)
+
+
 @pytest.mark.parametrize("enable_nightly_purge", [True])
 def test_auto_purge_disabled(hass_recorder):
     """Test periodic db cleanup still run when auto purge is disabled."""

tests/components/recorder/test_util.py

@@ -1,5 +1,5 @@
 """Test util methods."""
-from datetime import timedelta
+from datetime import datetime, timedelta
 import os
 import sqlite3
 from unittest.mock import MagicMock, patch
@@ -12,7 +12,11 @@ from homeassistant.components import recorder
 from homeassistant.components.recorder import run_information_with_session, util
 from homeassistant.components.recorder.const import DATA_INSTANCE, SQLITE_URL_PREFIX
 from homeassistant.components.recorder.models import RecorderRuns
-from homeassistant.components.recorder.util import end_incomplete_runs, session_scope
+from homeassistant.components.recorder.util import (
+    end_incomplete_runs,
+    is_second_sunday,
+    session_scope,
+)
 from homeassistant.util import dt as dt_util
 
 from .common import corrupt_db_file
@@ -584,3 +588,14 @@ async def test_write_lock_db(hass, tmp_path):
             # would be allowed to proceed as the goal is to prevent
             # all the other threads from accessing the database
             await hass.async_add_executor_job(_drop_table)
+
+
+def test_is_second_sunday():
+    """Test we can find the second sunday of the month."""
+    assert is_second_sunday(datetime(2022, 1, 9, 0, 0, 0, tzinfo=dt_util.UTC)) is True
+    assert is_second_sunday(datetime(2022, 2, 13, 0, 0, 0, tzinfo=dt_util.UTC)) is True
+    assert is_second_sunday(datetime(2022, 3, 13, 0, 0, 0, tzinfo=dt_util.UTC)) is True
+    assert is_second_sunday(datetime(2022, 4, 10, 0, 0, 0, tzinfo=dt_util.UTC)) is True
+    assert is_second_sunday(datetime(2022, 5, 8, 0, 0, 0, tzinfo=dt_util.UTC)) is True
+
+    assert is_second_sunday(datetime(2022, 1, 10, 0, 0, 0, tzinfo=dt_util.UTC)) is False