[tor-commits] [sbws/master] Merge branch 'ticket28565_rebased'

[juga at torproject.org]

commit f9df7658807b51434b93570add232b3becd0797f
Merge: 3db367b 127cf46
Author: juga0 <juga at riseup.net>
Date:   Thu Mar 21 12:46:01 2019 +0000

    Merge branch 'ticket28565_rebased'

 sbws/lib/v3bwfile.py            | 130 +++++++++++++++++++++++++++++++---------
 tests/unit/lib/test_v3bwfile.py |  44 +++++++++++---
 2 files changed, 139 insertions(+), 35 deletions(-)

diff --cc sbws/lib/v3bwfile.py
index 61657b9,e6dcde1..6a433d0
--- a/sbws/lib/v3bwfile.py
+++ b/sbws/lib/v3bwfile.py
@@@ -67,19 -89,7 +89,9 @@@ BW_HEADER_KEYVALUES_MONITOR = 
      # something else we don't know yet
      # So far is the number of ResultError
      'recent_measurement_failure_count',
-     # The number of success results should be:
-     # the number of attempts - the number of failures
-     # 4.6 header: the number of successful results, created in the last 5 days,
-     # that were excluded by a filter
-     # This is the sum of the following 3 + not success results
-     # 'recent_measurement_exclusion_count',
-     'recent_measurement_exclusion_not_distanciated_count',
-     'recent_measurement_exclusion_not_recent_count',
-     'recent_measurement_exclusion_not_min_num_count',
- 
 +    # The time it took to report about half of the network.
 +    'time_to_report_half_network',
- ]
+ ] + BW_HEADER_KEYVALUES_RECENT_MEASUREMENTS_EXCLUDED
  BANDWIDTH_HEADER_KEY_VALUES_INIT = \
      ['earliest_bandwidth', 'generator_started',
       'scanner_country', 'destinations_countries']\
@@@ -435,59 -453,15 +455,68 @@@ class V3BWHeader(object)
          [setattr(self, k, str(v)) for k, v in kwargs.items()
           if k in STATS_KEYVALUES]
  
 +    def add_time_report_half_network(self):
 +        """Add to the header the time it took to measure half of the network.
 +
 +        It is not the time the scanner actually takes on measuring all the
 +        network, but the ``number_eligible_relays`` that are reported in the
 +        bandwidth file and directory authorities will vote on.
 +
 +        This is calculated for half of the network, so that failed or not
 +        reported relays do not affect too much.
 +
 +        For instance, if there are 6500 relays in the network, half of the
 +        network would be 3250. And if there were 4000 eligible relays
 +        measured in an interval of 3 days, the time to measure half of the
 +        network would be 3 days * 3250 / 4000.
 +
 +        Since the elapsed time is calculated from the earliest and the
 +        latest measurement and a relay might have more than 2 measurements,
 +        this would give an estimate on how long it would take to measure
 +        the network including all the valid measurements.
 +
 +        Log also an estimated on how long it would take with the current
 +        number of relays included in the bandwidth file.
 +        """
 +        # NOTE: in future refactor do not convert attributes to str until
 +        # writing to the file, so that they do not need to be converted back
 +        # to do some calculations.
 +        elapsed_time = (
 +            (isostr_to_dt_obj(self.latest_bandwidth)
 +             - isostr_to_dt_obj(self.earliest_bandwidth))
 +            .total_seconds())
 +
 +        # This attributes were added later and some tests that
 +        # do not initialize them would fail.
 +        eligible_relays = int(getattr(self, 'number_eligible_relays', 0))
 +        consensus_relays = int(getattr(self, 'number_consensus_relays', 0))
 +        if not(eligible_relays and consensus_relays):
 +            return
 +
 +        half_network = consensus_relays / 2
 +        # Calculate the time it would take to measure half of the network
 +        if eligible_relays >= half_network:
 +            time_half_network = round(
 +                elapsed_time * half_network / eligible_relays
 +            )
 +            self.time_to_report_half_network = str(time_half_network)
 +
 +        # In any case log an estimated on the time to measure all the network.
 +        estimated_time = round(
 +            elapsed_time * consensus_relays / eligible_relays
 +        )
 +        log.info("Estimated time to measure the network: %s hours.",
 +                 round(estimated_time / 60 / 60))
 +
+     def add_relays_excluded_counters(self, exclusion_dict):
+         """
+         Add the monitoring KeyValues to the header about the number of
+         relays not included because they were not ``eligible``.
+         """
+         log.debug("Adding relays excluded counters.")
+         for k, v in exclusion_dict.items():
+             setattr(self, k, str(v))
+ 
  
  class V3BWLine(object):
      """