def test_metrics_stats_can_add_values():
stats = MetricsStats()
min_ts = trunc_ts(datetime.utcnow(), stats.timestep)
next_ts = min_ts + stats.timestep
gap_ts = next_ts + 3 * stats.timestep
max_ts = gap_ts + stats.timestep
first_events = [ev for ev in generate_events(10, min_ts) if ev.timestamp < min_ts + stats.timestep]
next_events = [ev for ev in generate_events(10, next_ts) if ev.timestamp < next_ts + stats.timestep]
after_gap_events = [ev for ev in generate_events(10, gap_ts) if ev.timestamp < gap_ts + stats.timestep]
all_events = first_events + next_events + after_gap_events
for ev in all_events:
stats.add(ev.timestamp, ev.name, ev.value)
assert stats.min_ts == min_ts
assert stats.max_ts == max_ts
assert sorted(ts for ts, _ in stats.frames()) == [min_ts, next_ts, gap_ts]
assert stats.frame(min_ts) == _metrics_stats_frame(first_events)
assert stats.frame(next_ts) == _metrics_stats_frame(next_events)
assert stats.frame(gap_ts) == _metrics_stats_frame(after_gap_events)
assert stats.frame(max_ts) == MetricsStatsFrame()
assert stats.frame(min_ts - 0.2 * stats.timestep) == MetricsStatsFrame()
assert stats.frame(min_ts + 0.8 * stats.timestep) == _metrics_stats_frame(first_events)
assert stats.frame(next_ts + 1.1 * stats.timestep) == MetricsStatsFrame()
assert stats.frame(gap_ts - 0.1 * stats.timestep) == MetricsStatsFrame()
assert stats.frame(max_ts + 4.2 * stats.timestep) == MetricsStatsFrame()
def test_metrics_stats_total_is_merge_of_all_frames():
events = generate_events(50)
stats = MetricsStats()
for ev in events:
stats.add(ev.timestamp, ev.name, ev.value)
expected_total = MetricsStatsFrame()
for _, frame in stats.frames():
expected_total.merge(frame)
assert stats.total == expected_total
def test_metrics_stats_frame_can_add_values():
events_transport_batch_size = [10, 2, 54]
events_looper_run_time_spent = [0.1, 3.4, 0.01, 0.5]
events = []
for v in events_transport_batch_size:
events.append((MetricsName.TRANSPORT_BATCH_SIZE, v))
for v in events_looper_run_time_spent:
events.append((MetricsName.LOOPER_RUN_TIME_SPENT, v))
shuffle(events)
frame = MetricsStatsFrame()
for id, value in events:
frame.add(id, value)
assert frame.get(MetricsName.TRANSPORT_BATCH_SIZE) == ValueAccumulator(events_transport_batch_size)
assert frame.get(MetricsName.LOOPER_RUN_TIME_SPENT) == ValueAccumulator(events_looper_run_time_spent)
assert frame.get(MetricsName.BACKUP_THREE_PC_BATCH_SIZE) == ValueAccumulator()
def test_metrics_stats_frame_eq_has_value_semantics():
a = MetricsStatsFrame()
b = MetricsStatsFrame()
assert a == b
a.add(MetricsName.LOOPER_RUN_TIME_SPENT, 2.0)
assert a != b
b.add(MetricsName.LOOPER_RUN_TIME_SPENT, 2.0)
assert a == b
a.add(MetricsName.BACKUP_THREE_PC_BATCH_SIZE, 1)
b.add(MetricsName.TRANSPORT_BATCH_SIZE, 2)
assert a != b
def _metrics_stats_frame(events):
frame = MetricsStatsFrame()
for ev in events:
frame.add(ev.name, ev.value)
return frame