def stop_start_job(self, frequency, start=datetime.utcnow(), restart=True):
"""
Restarts the job for a particular frequency.
"""
try:
logger.info("Stopping/starting %s at %s" % (frequency, start))
# Set up connection to W3ACT:
w = w3act(cfg.get('act','url'),cfg.get('act','username'),cfg.get('act','password'))
# Set up connection to H3:
h = hapyx.HapyX("https://%s:%s" % (cfg.get('h3','host'), cfg.get('h3','port')), username=cfg.get('h3','username'), password=cfg.get('h3','password'))
# Stop job if currently running:
if frequency in h.list_jobs() and h.status(frequency) != "":
"""Stops a running job, notifies RabbitMQ and cleans up the directory."""
launch_id = h.get_launch_id(frequency)
job = W3actJob.from_directory(w, "%s/%s" % (HERITRIX_JOBS, frequency), heritrix=h)
job.stop()
remove_action_files(frequency)
crawl.status.update_job_status.delay(job.name, "%s/%s" % (job.name, launch_id), "STOPPED")
# Pass on to the next step in the chain:
logger.info("Requesting assembly of output for: %s/%s" % (frequency, launch_id))
assemble_job_output.delay(frequency,launch_id)
else:
job = None
# Start job if requested:
if restart:
targets = w.get_ld_export(frequency)
# logger.info("Found %s Targets in export." % len(export))
# targets = [t for t in export if (t["startDate"] is None or t["startDate"] < start) and (t["endDateISO"] is None or t["crawlEndDateISO"] > start)]
logger.debug("Found %s Targets in date range." % len(targets))
job = W3actJob(w, targets, frequency, heritrix=h)
logger.info("Starting job %s..." % job.name)
job.start()
launch_id = h.get_launch_id(frequency)
crawl.status.update_job_status.delay(job.name, "%s/%s" % (job.name, launch_id), "LAUNCHED" )
logger.info("Launched job %s/%s with %s seeds." % (job.name, launch_id, len(job.seeds)))
return "Launched job %s/%s with %s seeds." % (job.name, launch_id, len(job.seeds))
else:
if job:
logger.info("Stopped job %s/%s without restarting..." % (job.name, launch_id))
return "Stopped job %s/%s without restarting..." % (job.name, launch_id)
else:
logger.warning("No running '%s' job to stop!" % frequency)
return "No running '%s' job to stop!" % frequency
except BaseException as e:
logger.exception(e)
raise self.retry(countdown=10, exe=e)
def run(self):
# Set up connection to H3:
h = get_hapy_for_job(self.job)
logger.info("I'm stopping %s" % (self.job.name))
# Stop job if currently running:
if self.job.name in h.list_jobs() and h.status(self.job.name) != "":
"""Stops a running job, cleans up the directory, initiates job assembly."""
launch_id = h.get_launch_id(self.job.name)
job = W3actJob.from_directory("%s/%s" % (h3().local_job_folder, self.job.name), heritrix=h)
job.stop()
remove_action_files(self.job.name, HERITRIX_JOBS=h3().local_job_folder)
# Record an output file that can be use as a Target by a different task:
mark_job_as(job, launch_id, 'stopped')
else:
logger.warning("No {} job to be stopped!".format(self.job.name))
def stop_start_job(self, frequency, start=datetime.utcnow(), restart=True):
"""
Restarts the job for a particular frequency.
"""
try:
logger.info("Stopping/starting %s at %s" % (frequency, start))
# Set up connection to W3ACT:
w = w3act(cfg.get('act', 'url'), cfg.get('act', 'username'),
cfg.get('act', 'password'))
# Set up connection to H3:
h = hapyx.HapyX("https://%s:%s" %
(cfg.get('h3', 'host'), cfg.get('h3', 'port')),
username=cfg.get('h3', 'username'),
password=cfg.get('h3', 'password'))
# Stop job if currently running:
if frequency in h.list_jobs() and h.status(frequency) != "":
"""Stops a running job, notifies RabbitMQ and cleans up the directory."""
launch_id = h.get_launch_id(frequency)
job = W3actJob.from_directory(w,
"%s/%s" % (HERITRIX_JOBS, frequency),
heritrix=h)
job.stop()
remove_action_files(frequency)
crawl.status.update_job_status.delay(
job.name, "%s/%s" % (job.name, launch_id), "STOPPED")
# Pass on to the next step in the chain:
logger.info("Requesting assembly of output for: %s/%s" %
(frequency, launch_id))
assemble_job_output.delay(frequency, launch_id)
else:
job = None
# Start job if requested:
if restart:
targets = w.get_ld_export(frequency)
# logger.info("Found %s Targets in export." % len(export))
# targets = [t for t in export if (t["startDate"] is None or t["startDate"] < start) and (t["endDateISO"] is None or t["crawlEndDateISO"] > start)]
logger.debug("Found %s Targets in date range." % len(targets))
job = W3actJob(w, targets, frequency, heritrix=h)
logger.info("Starting job %s..." % job.name)
job.start()
launch_id = h.get_launch_id(frequency)
crawl.status.update_job_status.delay(
job.name, "%s/%s" % (job.name, launch_id), "LAUNCHED")
logger.info("Launched job %s/%s with %s seeds." %
(job.name, launch_id, len(job.seeds)))
return "Launched job %s/%s with %s seeds." % (job.name, launch_id,
len(job.seeds))
else:
if job:
logger.info("Stopped job %s/%s without restarting..." %
(job.name, launch_id))
return "Stopped job %s/%s without restarting..." % (job.name,
launch_id)
else:
logger.warning("No running '%s' job to stop!" % frequency)
return "No running '%s' job to stop!" % frequency
except BaseException as e:
logger.exception(e)
raise self.retry(countdown=10, exe=e)
