commit 89801c5331b80776cced97d9da3a2990516279b3 Author: juga0 juga@riseup.net Date: Sat Mar 21 14:01:15 2020 +0000

chg: relaylist: Count measurements with timestamps

in RelayList: - Rename recent_measurement_attempt_count to recent_measurement_attempt when there is no counting - Use the timestamps class to manage/count measurement attempts --- sbws/core/scanner.py | 2 +- sbws/globals.py | 16 +++++++++++++++ sbws/lib/relaylist.py | 20 ++++++++++--------- sbws/lib/v3bwfile.py | 2 +- tests/unit/lib/test_relaylist.py | 42 ++++++++++++++++++++++++++++++++++++++++ tests/unit/lib/test_v3bwfile.py | 11 +++++++++++ 6 files changed, 82 insertions(+), 11 deletions(-)

diff --git a/sbws/core/scanner.py b/sbws/core/scanner.py index 60c8ca4..06381b8 100644 --- a/sbws/core/scanner.py +++ b/sbws/core/scanner.py @@ -515,7 +515,7 @@ def main_loop(args, conf, controller, relay_list, circuit_builder, result_dump, # Don't start measuring a relay if sbws is stopping. if settings.end_event.is_set(): break - relay_list.increment_recent_measurement_attempt_count() + relay_list.increment_recent_measurement_attempt() target.increment_relay_recent_measurement_attempt_count() num_relays += 1 # callback and callback_err must be non-blocking diff --git a/sbws/globals.py b/sbws/globals.py index a9dca10..01a147b 100644 --- a/sbws/globals.py +++ b/sbws/globals.py @@ -151,6 +151,22 @@ FACTOR_INCREMENT_DESTINATION_RETRY = 2 # Constants to check health KeyValues in the bandwidth file PERIOD_DAYS = int(MEASUREMENTS_PERIOD / (24 * 60 * 60)) MAX_RECENT_CONSENSUS_COUNT = PERIOD_DAYS * 24 # 120 +# XXX: This was only defined in `config.default.ini`, it should be read from +# here. +FRACTION_RELAYS = 0.05 +# A priority list currently takes more than 3h, ideally it should only take 1h. +MIN_HOURS_PRIORITY_LIST = 1 +# As of 2020, there're less than 7000 relays. +MAX_RELAYS = 8000 +# 120 +MAX_RECENT_PRIORITY_LIST_COUNT = int( + PERIOD_DAYS * 24 / MIN_HOURS_PRIORITY_LIST +) +MAX_RELAYS_PER_PRIORITY_LIST = int(MAX_RELAYS * FRACTION_RELAYS) # 400 +# 48000 +MAX_RECENT_PRIORITY_RELAY_COUNT = ( + MAX_RECENT_PRIORITY_LIST_COUNT * MAX_RELAYS_PER_PRIORITY_LIST +)

def fail_hard(*a, **kw): diff --git a/sbws/lib/relaylist.py b/sbws/lib/relaylist.py index 5ae1c2d..d5e5754 100644 --- a/sbws/lib/relaylist.py +++ b/sbws/lib/relaylist.py @@ -10,6 +10,7 @@ from threading import Lock

from ..globals import ( MAX_RECENT_CONSENSUS_COUNT, + MAX_RECENT_PRIORITY_RELAY_COUNT, MEASUREMENTS_PERIOD ) from ..util import timestamp, timestamps @@ -334,12 +335,10 @@ class RelayList: self._relays = [] # The period of time for which the measurements are keep. self._measurements_period = measurements_period - self._state = state - # NOTE: blocking: writes to disk - if self._state: - if self._state.get('recent_measurement_attempt_count', None) \ - is None: - self._state['recent_measurement_attempt_count'] = 0 + self._recent_measurement_attempt = timestamps.DateTimeSeq( + [], MAX_RECENT_PRIORITY_RELAY_COUNT, state, + "recent_measurement_attempt" + ) self._refresh()

def _need_refresh(self): @@ -502,7 +501,7 @@ class RelayList: return [r for r in self.exits if r.is_exit_not_bad_allowing_port(port)]

- def increment_recent_measurement_attempt_count(self): + def increment_recent_measurement_attempt(self): """ Increment the number of times that any relay has been queued to be measured. @@ -512,5 +511,8 @@ class RelayList: It is read and stored in a ``state`` file. """ # NOTE: blocking, writes to file! - if self._state: - self._state['recent_measurement_attempt_count'] += 1 + self._recent_measurement_attempt.update() + + @property + def recent_measurement_attempt_count(self): + return len(self._recent_measurement_attempt) diff --git a/sbws/lib/v3bwfile.py b/sbws/lib/v3bwfile.py index 70304a9..f81ef82 100644 --- a/sbws/lib/v3bwfile.py +++ b/sbws/lib/v3bwfile.py @@ -470,7 +470,7 @@ class V3BWHeader(object): in the recent (by default 5) days from the state file. """ state = State(state_fpath) - return state.get('recent_measurement_attempt_count', None) + return state.count('recent_measurement_attempt')

@staticmethod def recent_priority_list_count_from_file(state_fpath): diff --git a/tests/unit/lib/test_relaylist.py b/tests/unit/lib/test_relaylist.py index 1b22425..ea13b7b 100644 --- a/tests/unit/lib/test_relaylist.py +++ b/tests/unit/lib/test_relaylist.py @@ -75,3 +75,45 @@ def test_init_relays( # The number of relays will be the number of relays in the cosensus plus # the added ones minus the removed ones. assert 6925 == 6505 + len(added_fps) - len(removed_fps) + + +def test_increment_recent_measurement_attempt(args, conf, controller): + """Test that incrementing the measurement attempts does not go on forever + + And instead it only counts the number of attempts in the last days. + It also tests that the state file is updated correctly. + """ + state = State(conf['paths']['state_fpath']) + # For this test it does not matter that the consensus timestamps or relays + # are not correct. + relay_list = RelayList(args, conf, controller=controller, state=state) + # The initial count is 0 and the state does not have that key. + assert 0 == relay_list.recent_measurement_attempt_count + assert not state.get("recent_measurement_attempt", None) + + # Pretend that a measurement attempt is made. + with freeze_time("2020-02-29 10:00:00"): + relay_list.increment_recent_measurement_attempt() + assert 1 == relay_list.recent_measurement_attempt_count + assert [datetime(2020, 2, 29, 10, 0)] == state[ + "recent_measurement_attempt" + ] + + # And a second measurement attempt is made 4 days later. + with freeze_time("2020-03-04 10:00:00"): + relay_list.increment_recent_measurement_attempt() + assert 2 == relay_list.recent_measurement_attempt_count + assert 2 == len(state["recent_measurement_attempt"]) + + # And a third measurement attempt is made 5 days later. + with freeze_time("2020-03-05 10:00:00"): + relay_list.increment_recent_measurement_attempt() + assert 3 == relay_list.recent_measurement_attempt_count + assert 3 == len(state["recent_measurement_attempt"]) + + # And a fourth measurement attempt is made 6 days later. The first one is + # now removed and not counted. + with freeze_time("2020-03-06 10:00:00"): + relay_list.increment_recent_measurement_attempt() + assert 3 == relay_list.recent_measurement_attempt_count + assert 3 == len(state["recent_measurement_attempt"]) diff --git a/tests/unit/lib/test_v3bwfile.py b/tests/unit/lib/test_v3bwfile.py index c9c0e93..2ce34ff 100644 --- a/tests/unit/lib/test_v3bwfile.py +++ b/tests/unit/lib/test_v3bwfile.py @@ -540,3 +540,14 @@ def test_recent_consensus_count(root_data_path, datadir): results = load_result_file(str(datadir.join("results.txt"))) header = V3BWHeader.from_results(results, '', '', state_fpath) assert "1" == header.recent_consensus_count + + +def test_recent_measurement_attempt_count(root_data_path, datadir): + state_fpath = os.path.join(root_data_path, '.sbws/state.dat') + assert 15 == V3BWHeader.recent_measurement_attempt_count_from_file( + state_fpath + ) + # `results` does not matter here, using them to not have an empty list. + results = load_result_file(str(datadir.join("results.txt"))) + header = V3BWHeader.from_results(results, '', '', state_fpath) + assert "15" == header.recent_measurement_attempt_count