def test_cache_miss_value_error_size_of_table():
"""
ValueError should be raised if we try to get the size of something not in cache.
"""
connection_mock = Mock()
connection_mock.fetch.return_value = []
with pytest.raises(ValueError):
get_size_of_table(connection_mock, "DUMMY_SCHEMA", "DUMMY_NAME")
def test_size_of_table(flowmachine_connect):
"""
Test that table size is reported correctly.
"""
dl = daily_location("2016-01-01").store().result()
total_cache_size = get_size_of_cache(flowmachine_connect)
table_size = get_size_of_table(flowmachine_connect, dl.table_name, "cache")
assert total_cache_size == table_size
def test_scoring(flowmachine_connect):
"""
Test that score updating algorithm is correct by comparing to cachey as reference implementation
"""
dl = daily_location("2016-01-01").store().result()
dl_time = get_compute_time(get_db(), dl.query_id)
dl_size = get_size_of_table(get_db(), dl.table_name, "cache")
initial_score = get_score(get_db(), dl.query_id)
cachey_scorer = Scorer(halflife=1000.0)
cache_score = cachey_scorer.touch("dl", dl_time / dl_size)
assert cache_score == pytest.approx(initial_score)
# Touch again
new_score = touch_cache(get_db(), dl.query_id)
updated_cache_score = cachey_scorer.touch("dl")
assert updated_cache_score == pytest.approx(new_score)
# Add another unrelated cache record, which should have a higher initial score
dl_2 = daily_location("2016-01-02").store().result()
dl_time = get_compute_time(get_db(), dl_2.query_id)
dl_size = get_size_of_table(get_db(), dl_2.table_name, "cache")
cache_score = cachey_scorer.touch("dl_2", dl_time / dl_size)
assert cache_score == pytest.approx(get_score(get_db(), dl_2.query_id))
def test_shrink_to_size_dry_run_reflects_wet_run(flowmachine_connect):
"""
Test that shrink_below_size dry run is an accurate report.
"""
dl = daily_location("2016-01-01").store().result()
dl2 = daily_location("2016-01-02").store().result()
shrink_to = get_size_of_table(flowmachine_connect, dl.table_name, "cache")
queries_that_would_be_removed = shrink_below_size(flowmachine_connect,
shrink_to,
dry_run=True)
removed_queries = shrink_below_size(flowmachine_connect,
shrink_to,
dry_run=False)
assert [q.md5 for q in removed_queries
] == [q.md5 for q in queries_that_would_be_removed]
def test_shrink_to_size_uses_score(flowmachine_connect):
"""
Test that shrink_below_size removes cache records in ascending score order.
"""
dl = daily_location("2016-01-01").store().result()
dl_aggregate = dl.aggregate().store().result()
flowmachine_connect.engine.execute(
f"UPDATE cache.cached SET cache_score_multiplier = 100 WHERE query_id='{dl_aggregate.md5}'"
)
flowmachine_connect.engine.execute(
f"UPDATE cache.cached SET cache_score_multiplier = 0.5 WHERE query_id='{dl.md5}'"
)
table_size = get_size_of_table(flowmachine_connect, dl.table_name, "cache")
removed_queries = shrink_below_size(flowmachine_connect, table_size)
assert 1 == len(removed_queries)
assert not dl.is_stored
assert dl_aggregate.is_stored
