mirror of
https://github.com/internetarchive/brozzler.git
synced 2025-02-24 08:39:59 -05:00
fix and test both job stop request and site stop requests
This commit is contained in:
parent
8aa10962bc
commit
bb98a43c8c
@ -264,6 +264,7 @@ class RethinkDbFrontier:
|
|||||||
# can't call resume_job since that would resume jobs's other sites
|
# can't call resume_job since that would resume jobs's other sites
|
||||||
job = brozzler.Job.load(self.rr, site.job_id)
|
job = brozzler.Job.load(self.rr, site.job_id)
|
||||||
job.status = "ACTIVE"
|
job.status = "ACTIVE"
|
||||||
|
site.stop_requested = None
|
||||||
job.starts_and_stops.append(
|
job.starts_and_stops.append(
|
||||||
{"start":doublethink.utcnow(), "stop":None})
|
{"start":doublethink.utcnow(), "stop":None})
|
||||||
job.save()
|
job.save()
|
||||||
|
@ -249,42 +249,104 @@ def test_resume_job():
|
|||||||
assert site.starts_and_stops[3]['start']
|
assert site.starts_and_stops[3]['start']
|
||||||
assert site.starts_and_stops[3]['stop'] is None
|
assert site.starts_and_stops[3]['stop'] is None
|
||||||
|
|
||||||
# simulate a crawl stopped by a stop request
|
# simulate a job stop request
|
||||||
|
job_conf = {'seeds': [{'url': 'http://example.com/'}, {'url': 'http://example_2.com/'}]}
|
||||||
|
job = brozzler.new_job(frontier, job_conf)
|
||||||
|
assert len(list(frontier.job_sites(job.id))) == 2
|
||||||
|
site1 = list(frontier.job_sites(job.id))[0]
|
||||||
|
site2 = list(frontier.job_sites(job.id))[1]
|
||||||
|
|
||||||
job.stop_requested = datetime.datetime.utcnow().replace(tzinfo=doublethink.UTC)
|
job.stop_requested = datetime.datetime.utcnow().replace(tzinfo=doublethink.UTC)
|
||||||
job.save()
|
job.save()
|
||||||
|
|
||||||
|
# should raise a CrawlStopped
|
||||||
with pytest.raises(brozzler.CrawlStopped):
|
with pytest.raises(brozzler.CrawlStopped):
|
||||||
frontier.honor_stop_request(site)
|
frontier.honor_stop_request(site1)
|
||||||
|
|
||||||
frontier.finished(site, 'FINISHED_STOP_REQUESTED')
|
frontier.finished(site1, 'FINISHED_STOP_REQUESTED')
|
||||||
|
frontier.finished(site2, 'FINISHED_STOP_REQUESTED')
|
||||||
job.refresh()
|
job.refresh()
|
||||||
|
|
||||||
assert job.status == 'FINISHED'
|
assert job.status == 'FINISHED'
|
||||||
assert job.stop_requested
|
assert job.stop_requested
|
||||||
assert len(job.starts_and_stops) == 4
|
assert len(job.starts_and_stops) == 1
|
||||||
assert job.starts_and_stops[3]['start']
|
assert job.starts_and_stops[0]['start']
|
||||||
assert job.starts_and_stops[3]['stop']
|
assert job.starts_and_stops[0]['stop']
|
||||||
assert job.starts_and_stops[3]['stop'] > job.starts_and_stops[3]['start']
|
assert job.starts_and_stops[0]['stop'] > job.starts_and_stops[0]['start']
|
||||||
assert site.status == 'FINISHED_STOP_REQUESTED'
|
assert site1.status == 'FINISHED_STOP_REQUESTED'
|
||||||
assert len(site.starts_and_stops) == 4
|
assert site2.status == 'FINISHED_STOP_REQUESTED'
|
||||||
assert site.starts_and_stops[3]['start']
|
assert len(site1.starts_and_stops) == 1
|
||||||
assert site.starts_and_stops[3]['stop']
|
assert len(site2.starts_and_stops) == 1
|
||||||
assert site.starts_and_stops[3]['stop'] > site.starts_and_stops[3]['start']
|
assert site1.starts_and_stops[0]['start']
|
||||||
|
assert site1.starts_and_stops[0]['stop']
|
||||||
|
assert site1.starts_and_stops[0]['stop'] > site.starts_and_stops[0]['start']
|
||||||
|
assert site2.starts_and_stops[0]['start']
|
||||||
|
assert site2.starts_and_stops[0]['stop']
|
||||||
|
assert site2.starts_and_stops[0]['stop'] > site.starts_and_stops[0]['start']
|
||||||
|
|
||||||
# test resume job after a stop request
|
# simulate job resume after a stop request
|
||||||
frontier.resume_job(job)
|
frontier.resume_job(job)
|
||||||
site = list(frontier.job_sites(job.id))[0]
|
site1 = list(frontier.job_sites(job.id))[0]
|
||||||
|
site2 = list(frontier.job_sites(job.id))[0]
|
||||||
|
|
||||||
assert job.status == 'ACTIVE'
|
assert job.status == 'ACTIVE'
|
||||||
assert job.stop_requested is None
|
assert job.stop_requested is None
|
||||||
assert len(job.starts_and_stops) == 5
|
assert len(job.starts_and_stops) == 2
|
||||||
assert job.starts_and_stops[4]['start']
|
assert job.starts_and_stops[1]['start']
|
||||||
assert job.starts_and_stops[4]['stop'] is None
|
assert job.starts_and_stops[1]['stop'] is None
|
||||||
assert site.status == 'ACTIVE'
|
assert site1.status == 'ACTIVE'
|
||||||
assert len(site.starts_and_stops) == 5
|
assert len(site1.starts_and_stops) == 2
|
||||||
assert site.starts_and_stops[4]['start']
|
assert site1.starts_and_stops[1]['start']
|
||||||
assert site.starts_and_stops[4]['stop'] is None
|
assert site1.starts_and_stops[1]['stop'] is None
|
||||||
|
assert site2.status == 'ACTIVE'
|
||||||
|
assert len(site2.starts_and_stops) == 2
|
||||||
|
assert site2.starts_and_stops[1]['start']
|
||||||
|
assert site2.starts_and_stops[1]['stop'] is None
|
||||||
|
|
||||||
|
# simulate a site stop request
|
||||||
|
site1.stop_requested = datetime.datetime.utcnow().replace(tzinfo=doublethink.UTC)
|
||||||
|
site1.save()
|
||||||
|
|
||||||
|
# should not raise a CrawlStopped
|
||||||
|
frontier.honor_stop_request(site)
|
||||||
|
|
||||||
|
frontier.finished(site1, 'FINISHED_STOP_REQUESTED')
|
||||||
|
job.refresh()
|
||||||
|
|
||||||
|
assert job.status == 'ACTIVE'
|
||||||
|
assert job.stop_requested is None
|
||||||
|
assert len(job.starts_and_stops) == 2
|
||||||
|
assert job.starts_and_stops[1]['start']
|
||||||
|
assert job.starts_and_stops[1]['stop'] is None
|
||||||
|
assert site1.status == 'FINISHED_STOP_REQUESTED'
|
||||||
|
assert len(site1.starts_and_stops) == 2
|
||||||
|
assert site1.starts_and_stops[1]['start']
|
||||||
|
assert site1.starts_and_stops[1]['stop']
|
||||||
|
assert site1.starts_and_stops[1]['stop'] > site.starts_and_stops[1]['start']
|
||||||
|
assert site2.status == 'ACTIVE'
|
||||||
|
assert len(site2.starts_and_stops) == 2
|
||||||
|
assert site2.starts_and_stops[1]['start']
|
||||||
|
assert site2.starts_and_stops[1]['stop'] is None
|
||||||
|
|
||||||
|
# simulate site resume after a stop request
|
||||||
|
frontier.resume_site(site1)
|
||||||
|
site1 = list(frontier.job_sites(job.id))[0]
|
||||||
|
site2 = list(frontier.job_sites(job.id))[1]
|
||||||
|
|
||||||
|
assert job.status == 'ACTIVE'
|
||||||
|
assert job.stop_requested is None
|
||||||
|
assert len(job.starts_and_stops) == 2
|
||||||
|
assert job.starts_and_stops[1]['start']
|
||||||
|
assert job.starts_and_stops[1]['stop'] is None
|
||||||
|
assert site1.status == 'ACTIVE'
|
||||||
|
assert site1.stop_requested is None
|
||||||
|
assert len(site1.starts_and_stops) == 3
|
||||||
|
assert site1.starts_and_stops[2]['start']
|
||||||
|
assert site1.starts_and_stops[2]['stop'] is None
|
||||||
|
assert site2.status == 'ACTIVE'
|
||||||
|
assert len(site2.starts_and_stops) == 2
|
||||||
|
assert site2.starts_and_stops[1]['start']
|
||||||
|
assert site2.starts_and_stops[1]['stop'] is None
|
||||||
|
|
||||||
def test_time_limit():
|
def test_time_limit():
|
||||||
# XXX test not thoroughly adapted to change in time accounting, since
|
# XXX test not thoroughly adapted to change in time accounting, since
|
||||||
|
Loading…
x
Reference in New Issue
Block a user