skip active_sites query

This commit is contained in:
Barbara Miller 2020-03-11 13:40:37 -07:00
parent c2a1ca018a
commit af39b8cc6f

View File

@ -80,17 +80,7 @@ class BrozzlerWorker:
warcproxes = self._service_registry.available_services('warcprox')
if not warcproxes:
return None
reql = self._frontier.rr.table('sites').between(
['ACTIVE', r.minval], ['ACTIVE', r.maxval],
index='sites_last_disclaimed')
active_sites = list(reql.run())
for warcprox in warcproxes:
address = '%s:%s' % (warcprox['host'], warcprox['port'])
warcprox['assigned_sites'] = len([
site for site in active_sites
if 'proxy' in site and site['proxy'] == address])
warcproxes.sort(key=lambda warcprox: (
warcprox['assigned_sites'], warcprox['load']))
warcproxes.sort(key=lambda warcprox: (warcprox['load']))
# XXX make this heuristic more advanced?
return warcproxes[0]