def testPurgeOldMetricDataRowsFewerCandidatesThanExpected(
self, estimateNumRowsToDeleteMock, queryCandidateRowsMock,
deleteRowsMock):
estimate = metric_garbage_collector._MAX_DELETE_BATCH_SIZE * 2
estimateNumRowsToDeleteMock.return_value = estimate
candidatesIter = iter(
zip(["ABCDEF"] * (estimate // 2), xrange(estimate // 2)))
queryCandidateRowsMock.side_effect = (lambda limit, **kwargs: tuple(
itertools.islice(candidatesIter, limit)))
deleteRowsMock.side_effect = (
lambda uidRowidPairs, **kwargs: len(uidRowidPairs))
# Execute
numDeleted = metric_garbage_collector.purgeOldMetricDataRows(
thresholdDays=90)
self.assertEqual(numDeleted, estimate // 2)
self.assertEqual(estimateNumRowsToDeleteMock.call_count, 1)
self.assertEqual(queryCandidateRowsMock.call_count, 2)
self.assertEqual(deleteRowsMock.call_count, 1)
def testPurgeOldMetricDataRowsDeletedLessThanExpected(
self, estimateNumRowsToDeleteMock, queryCandidateRowsMock,
deleteRowsMock):
estimate = metric_garbage_collector._MAX_DELETE_BATCH_SIZE * 3
estimateNumRowsToDeleteMock.return_value = estimate
candidatesIter = iter(zip(["ABCDEF"] * estimate, xrange(estimate)))
queryCandidateRowsMock.side_effect = (lambda limit, **kwargs: tuple(
itertools.islice(candidatesIter, limit)))
deletedCounts = [
metric_garbage_collector._MAX_DELETE_BATCH_SIZE,
metric_garbage_collector._MAX_DELETE_BATCH_SIZE // 2,
metric_garbage_collector._MAX_DELETE_BATCH_SIZE
]
deleteRowsMock.side_effect = iter(deletedCounts)
# Execute
numDeleted = metric_garbage_collector.purgeOldMetricDataRows(
thresholdDays=90)
self.assertEqual(numDeleted, sum(deletedCounts))
self.assertEqual(estimateNumRowsToDeleteMock.call_count, 1)
self.assertEqual(queryCandidateRowsMock.call_count, 4)
self.assertEqual(deleteRowsMock.call_count, 3)
def testPurgeOldMetricDataRowsStopAtEstimated(self,
estimateNumRowsToDeleteMock,
queryCandidateRowsMock,
deleteRowsMock):
estimate = metric_garbage_collector._MAX_DELETE_BATCH_SIZE * 2
estimateNumRowsToDeleteMock.return_value = estimate
candidatesIter = iter(
zip(["ABCDEF"] * (estimate + 1), xrange(estimate + 1)))
queryCandidateRowsMock.side_effect = (lambda limit, **kwargs: tuple(
itertools.islice(candidatesIter, limit)))
deleteRowsMock.side_effect = (
lambda uidRowidPairs, **kwargs: len(uidRowidPairs))
# Execute
numDeleted = metric_garbage_collector.purgeOldMetricDataRows(
thresholdDays=90)
self.assertEqual(numDeleted, estimate)
self.assertEqual(estimateNumRowsToDeleteMock.call_count, 1)
self.assertEqual(queryCandidateRowsMock.call_count, 2)
self.assertEqual(deleteRowsMock.call_count, 2)
# Make sure it didn't try to retrieve candidates beyond estimated number
self.assertEqual(len(tuple(candidatesIter)), 1)
def testPurgeOldMetricDataRowsDeletedLessThanExpected(
self, estimateNumRowsToDeleteMock, queryCandidateRowsMock, deleteRowsMock
):
estimate = metric_garbage_collector._MAX_DELETE_BATCH_SIZE * 3
estimateNumRowsToDeleteMock.return_value = estimate
candidatesIter = iter(zip(["ABCDEF"] * estimate, xrange(estimate)))
queryCandidateRowsMock.side_effect = lambda limit, **kwargs: tuple(itertools.islice(candidatesIter, limit))
deletedCounts = [
metric_garbage_collector._MAX_DELETE_BATCH_SIZE,
metric_garbage_collector._MAX_DELETE_BATCH_SIZE // 2,
metric_garbage_collector._MAX_DELETE_BATCH_SIZE,
]
deleteRowsMock.side_effect = iter(deletedCounts)
# Execute
numDeleted = metric_garbage_collector.purgeOldMetricDataRows(thresholdDays=90)
self.assertEqual(numDeleted, sum(deletedCounts))
self.assertEqual(estimateNumRowsToDeleteMock.call_count, 1)
self.assertEqual(queryCandidateRowsMock.call_count, 4)
self.assertEqual(deleteRowsMock.call_count, 3)
def testPurgeOldMetricDataRowsStopAtEstimated(
self, estimateNumRowsToDeleteMock, queryCandidateRowsMock, deleteRowsMock
):
estimate = metric_garbage_collector._MAX_DELETE_BATCH_SIZE * 2
estimateNumRowsToDeleteMock.return_value = estimate
candidatesIter = iter(zip(["ABCDEF"] * (estimate + 1), xrange(estimate + 1)))
queryCandidateRowsMock.side_effect = lambda limit, **kwargs: tuple(itertools.islice(candidatesIter, limit))
deleteRowsMock.side_effect = lambda uidRowidPairs, **kwargs: len(uidRowidPairs)
# Execute
numDeleted = metric_garbage_collector.purgeOldMetricDataRows(thresholdDays=90)
self.assertEqual(numDeleted, estimate)
self.assertEqual(estimateNumRowsToDeleteMock.call_count, 1)
self.assertEqual(queryCandidateRowsMock.call_count, 2)
self.assertEqual(deleteRowsMock.call_count, 2)
# Make sure it didn't try to retrieve candidates beyond estimated number
self.assertEqual(len(tuple(candidatesIter)), 1)
def testPurgeOldMetricDataRowsWithoutOldRecords(
self, estimateNumRowsToDeleteMock, queryCandidateRowsMock, deleteRowsMock
):
estimateNumRowsToDeleteMock.return_value = 0
# These should not be called in this test
queryCandidateRowsMock.side_effect = []
deleteRowsMock.side_effect = []
numDeleted = metric_garbage_collector.purgeOldMetricDataRows(thresholdDays=90)
self.assertEqual(numDeleted, 0)
self.assertEqual(estimateNumRowsToDeleteMock.call_count, 1)
self.assertEqual(queryCandidateRowsMock.call_count, 0)
self.assertEqual(deleteRowsMock.call_count, 0)
def testPurgeOldMetricDataRowsWithoutOldRecords(
self, estimateNumRowsToDeleteMock, queryCandidateRowsMock,
deleteRowsMock):
estimateNumRowsToDeleteMock.return_value = 0
# These should not be called in this test
queryCandidateRowsMock.side_effect = []
deleteRowsMock.side_effect = []
numDeleted = metric_garbage_collector.purgeOldMetricDataRows(
thresholdDays=90)
self.assertEqual(numDeleted, 0)
self.assertEqual(estimateNumRowsToDeleteMock.call_count, 1)
self.assertEqual(queryCandidateRowsMock.call_count, 0)
self.assertEqual(deleteRowsMock.call_count, 0)
def testPurgeOldMetricDataRowsFewerCandidatesThanExpected(
self, estimateNumRowsToDeleteMock, queryCandidateRowsMock, deleteRowsMock
):
estimate = metric_garbage_collector._MAX_DELETE_BATCH_SIZE * 2
estimateNumRowsToDeleteMock.return_value = estimate
candidatesIter = iter(zip(["ABCDEF"] * (estimate // 2), xrange(estimate // 2)))
queryCandidateRowsMock.side_effect = lambda limit, **kwargs: tuple(itertools.islice(candidatesIter, limit))
deleteRowsMock.side_effect = lambda uidRowidPairs, **kwargs: len(uidRowidPairs)
# Execute
numDeleted = metric_garbage_collector.purgeOldMetricDataRows(thresholdDays=90)
self.assertEqual(numDeleted, estimate // 2)
self.assertEqual(estimateNumRowsToDeleteMock.call_count, 1)
self.assertEqual(queryCandidateRowsMock.call_count, 2)
self.assertEqual(deleteRowsMock.call_count, 1)
def testPurgeOldMetricData(self):
gcThresholdDays = 90
now = datetime.utcnow().replace(microsecond=0)
uid1 = uuid.uuid1().hex
oldRows = [
dict(
value=1.0,
timestamp=now - timedelta(days=gcThresholdDays + 1),
),
dict(
value=2.0,
timestamp=now - timedelta(days=gcThresholdDays + 2),
),
]
youngRows = [
dict(
value=3.0,
timestamp=now,
),
dict(
value=4.0,
timestamp=now - timedelta(days=gcThresholdDays - 1),
),
dict(
value=5.0,
timestamp=now - timedelta(days=gcThresholdDays - 2),
),
]
allRows = oldRows + youngRows
# Use a temporary database
with repository_test_utils.HtmengineManagedTempRepository("metric_gc"):
engine = htmengine.repository.engineFactory(
config=htmengine.APP_CONFIG)
# Add the dummy metric rows
allData = [(row["value"], row["timestamp"]) for row in allRows]
with engine.connect() as conn: # pylint: disable=E1101
htmengine.repository.addMetric(conn, uid=uid1)
insertedObjects = htmengine.repository.addMetricData(
conn, metricId=uid1, data=allData)
numInserted = len(insertedObjects)
self.assertEqual(numInserted, len(allRows))
# Execute
numDeleted = metric_garbage_collector.purgeOldMetricDataRows(
gcThresholdDays)
# Verify
self.assertEqual(numDeleted, len(oldRows))
# Verify that only the old tweets got purged
with engine.connect() as conn: # pylint: disable=E1101
remainingRows = htmengine.repository.getMetricData(
conn).fetchall()
self.assertEqual(len(remainingRows), len(youngRows))
self.assertItemsEqual([(row["value"], row["timestamp"])
for row in youngRows],
[(row.metric_value, row.timestamp)
for row in remainingRows]) # pylint: disable=E1101
def testPurgeOldMetricData(self):
gcThresholdDays = 90
now = datetime.utcnow().replace(microsecond=0)
uid1 = uuid.uuid1().hex
oldRows = [
dict(
value=1.0,
timestamp=now - timedelta(days=gcThresholdDays + 1),
),
dict(
value=2.0,
timestamp=now - timedelta(days=gcThresholdDays + 2),
),
]
youngRows = [
dict(
value=3.0,
timestamp=now,
),
dict(
value=4.0,
timestamp=now - timedelta(days=gcThresholdDays - 1),
),
dict(
value=5.0,
timestamp=now - timedelta(days=gcThresholdDays - 2),
),
]
allRows = oldRows + youngRows
# Use a temporary database
with repository_test_utils.ManagedTempRepository("metric_gc"):
engine = htmengine.repository.engineFactory(config=htmengine.APP_CONFIG)
# Add the dummy metric rows
allData = [(row["value"], row["timestamp"]) for row in allRows]
with engine.connect() as conn: # pylint: disable=E1101
htmengine.repository.addMetric(conn, uid=uid1)
insertedObjects = htmengine.repository.addMetricData(conn,
metricId=uid1,
data=allData)
numInserted = len(insertedObjects)
self.assertEqual(numInserted, len(allRows))
# Execute
numDeleted = metric_garbage_collector.purgeOldMetricDataRows(
gcThresholdDays)
# Verify
self.assertEqual(numDeleted, len(oldRows))
# Verify that only the old tweets got purged
with engine.connect() as conn: # pylint: disable=E1101
remainingRows = htmengine.repository.getMetricData(conn).fetchall()
self.assertEqual(len(remainingRows), len(youngRows))
self.assertItemsEqual(
[(row["value"], row["timestamp"]) for row in youngRows],
[(row.metric_value, row.timestamp) for row in remainingRows]) # pylint: disable=E1101
