def process_message(self, pid, method):
# process an update message from fedora
# when an object is purged from fedora, remove it from the index
if method == 'purgeObject':
# since we don't know which index (if any) this object was indexed in,
# delete it from all configured indexes
for site, index in self.indexes.iteritems():
try:
index.delete_item(pid)
except Exception as e:
logging.error("Failed to purge %s (%s): %s",
pid, site, e)
# Add a prefix to the detail error message if we
# can identify what type of error this is.
detail_type = ''
if isinstance(e, SolrError):
detail_type = 'Solr Error: '
action_str = 'Purge: '
msg = '%s%s%s' % (detail_type, action_str, e)
err = IndexError(object_id=pid, site=site, detail=msg)
err.save()
logger.info('Deleting %s from all configured Solr indexes', pid)
# commit?
# ingest, modify object or modify datastream
else:
# if the object isn't already in the queue to be indexed, check if it should be
if pid not in self.to_index:
# get content models from resource index
obj_cmodels = list(self.repo.risearch.get_objects('info:fedora/%s' % pid,
modelns.hasModel))
sample_obj = self.repo.get_object(pid)
obj_cmodels2 = sample_obj.get_models()
logger.debug('Logging object cmodels %s', modelns.hasModel)
logger.debug('Logging object cmodels %s', obj_cmodels)
logger.debug('Logging object cmodels %s', obj_cmodels2)
# may include generic content models, but should not be a problem
# find which configured site(s) index the item
for site, index in self.indexes.iteritems():
if index.indexes_item(obj_cmodels):
if pid not in self.to_index:
# first site found - create a queue item and add to the list
self.to_index[pid] = QueueItem(site)
else:
# subsequent site - add the site to the existing queue item
self.to_index[pid].add_site(site)
def reindex_object(self, site, pid):
indexes = load_indexes()
site_index = indexes[site]
try:
indexed = site_index.index_item(pid)
err = None
except Exception as e:
logging.error("Failed to index %s (%s): %s",
pid, site, e)
# Add a prefix to the detail error message if we
# can identify what type of error this is.
detail_type = ''
if isinstance(e, SolrError):
detail_type = 'Solr Error: '
msg = '%s%s' % (detail_type, e)
err = IndexError(object_id=pid, site=site,
detail=msg)
err.save()
return 'Indexed pid %s' % pid
def index_item(self, pid, queueitem, site):
'''Index an item in a single configured site index and handle
any errors, updating the queueitem retry count and marking
sites as indexed according to success or any errors.
:param pid: pid for the item to be indexed
:param queueitem: :class:`QueueItem`
:param site: name of the site index to use
'''
try:
# tell the site index to index the item - returns True on success
if self.indexes[site].index_item(pid):
# mark the site index as complete on the queued item
self.to_index[pid].site_complete(site)
except RecoverableIndexError as rie:
# If the index attempt resulted in error that we
# can potentially recover from, keep the item in
# the queue and attempt to index it again.
# Increase the count of index attempts, so we know when to stop.
self.to_index[pid].tries += 1
# quit when we reached the configured number of index attempts
if self.to_index[pid].tries >= self.index_max_tries:
logger.error("Failed to index %s (%s) after %d tries: %s",
pid, site, self.to_index[pid].tries, rie)
err = IndexError(object_id=pid, site=site,
detail='Failed to index after %d attempts: %s' % \
(self.to_index[pid].tries, rie))
err.save()
# we've hit the index retry limit, so set site as complete on the queue item
self.to_index[pid].site_complete(site)
else:
logging.warn("Recoverable error attempting to index %s (%s), %d tries: %s",
pid, site, self.to_index[pid].tries, rie)
# update the index time - wait the configured index delay before
# attempting to reindex again
self.to_index[pid].time = datetime.now()
except Exception as e:
logging.error("Failed to index %s (%s): %s",
pid, site, e)
# Add a prefix to the detail error message if we
# can identify what type of error this is.
detail_type = ''
if isinstance(e, SolrError):
detail_type = 'Solr Error: '
msg = '%s%s' % (detail_type, e)
err = IndexError(object_id=pid, site=site,
detail=msg)
err.save()
# any exception not caught in the recoverable error block
# should not be attempted again - set site as complete on queue item
self.to_index[pid].site_complete(site)
def index_object(self, pid, site):
indexes = load_indexes()
index_max_tries = 3
try:
# tell the site index to index the item
indexes[site].index_item(pid)
except RecoverableIndexError as rie:
# If the index attempt resulted in error that we
# can potentially recover from, keep the item in
# the queue and attempt to index it again.
self.retry(countdown=2, exc=rie)
if index_object.request.retries >= index_object.max_retries:
logger.error("Failed to index %s (%s) after %d tries: %s",
pid, site, index_object.request.retries, rie)
err = IndexError(object_id=pid, site=site,
detail='Failed to index after %d attempts: %s' % \
(index_object.request.retries, rie))
err.save()
else:
logging.warn("Recoverable error attempting to index %s (%s), %d tries: %s",
pid, site, index_object.request.retries, rie)
except Exception as e:
logging.error("Failed to index %s (%s): %s",
pid, site, e)
# Add a prefix to the detail error message if we
# can identify what type of error this is.
detail_type = ''
if isinstance(e, SolrError):
detail_type = 'Solr Error: '
msg = '%s%s' % (detail_type, e)
err = IndexError(object_id=pid, site=site,
detail=msg)
err.save()
# any exception not caught in the recoverable error block
# should not be attempted again - set site as complete on queue item
# self.to_index[pid].site_complete(site)
return 'Indexed pid %s' % pid