def _case_to_case_index_ratio(self):
if not should_use_sql_backend(self.domain):
self.stdout.write('\nUnable to get case to index ratio of Couch domain\n')
return
db_name = get_db_aliases_for_partitioned_query()[0] # just query one shard DB
case_query = CommCareCaseSQL.objects.using(db_name).filter(domain=self.domain)
index_query = CommCareCaseIndexSQL.objects.using(db_name).filter(domain=self.domain)
case_count = estimate_row_count(case_query, db_name)
case_index_count = estimate_row_count(index_query, db_name)
self._print_value('Ratio of cases to case indices: 1 : ', case_index_count / case_count)
def _case_to_case_index_ratio(self):
db_name = get_db_aliases_for_partitioned_query()[
0] # just query one shard DB
case_query = CommCareCase.objects.using(db_name).filter(
domain=self.domain)
index_query = CommCareCaseIndex.objects.using(db_name).filter(
domain=self.domain)
case_count = estimate_row_count(case_query, db_name)
case_index_count = estimate_row_count(index_query, db_name)
self._print_value('Ratio of cases to case indices: 1 : ',
case_index_count / case_count)
def get_approximate_doc_count(self, from_db):
"""Get the approximate doc count from the given DB
:param from_db: The DB alias to query
"""
query = self.query(from_db, for_count=True)
return estimate_row_count(query, from_db)
def _get_count(config):
table_name = get_table_name(config.domain, config.table_id)
db_name = connection_manager.get_django_db_alias(config.engine_id)
query = ('SELECT * FROM "%s"' % table_name, [])
try:
return estimate_row_count(query, db_name)
except ProgrammingError:
return "Table not found"
def get_sql_repeat_record_count(domain, repeater_id=None, state=None):
from .models import SQLRepeatRecord
queryset = SQLRepeatRecord.objects.filter(domain=domain)
if repeater_id:
queryset = queryset.filter(repeater__repeater_id=repeater_id)
if state:
queryset = queryset.filter(state=state)
return estimate_row_count(queryset)
def delete_old_images_on_db(db_name, cutoff):
if isinstance(cutoff, str):
cutoff = parse_date(cutoff, default_timezone=None)
max_age = cutoff - timedelta(days=90)
db = get_blob_db()
def _get_query(db_name, max_age=max_age):
return BlobMeta.objects.using(db_name).filter(
type_code=CODES.form_attachment,
domain='icds-cas',
created_on__lt=max_age).order_by('created_on')
bytes_deleted = 0
query = _get_query(db_name)
metas = list(query[:1000])
run_again = len(metas) == 1000
if metas:
for meta in metas:
bytes_deleted += meta.content_length or 0
db.bulk_delete(metas=metas)
tags = {'database': db_name}
age = datetime.utcnow() - metas[-1].created_on
metrics_gauge('commcare.icds_images.max_age',
value=age.total_seconds(),
tags=tags)
row_estimate = estimate_row_count(query, db_name)
metrics_gauge('commcare.icds_images.count_estimate',
value=row_estimate,
tags=tags)
metrics_counter('commcare.icds_images.bytes_deleted',
value=bytes_deleted)
metrics_counter('commcare.icds_images.count_deleted', value=len(metas))
runtime = datetime.utcnow() - cutoff
if run_again and runtime.total_seconds() < MAX_RUNTIME:
delete_old_images_on_db.delay(db_name, cutoff)
def _get_count(config):
table_name = get_table_name(config.domain, config.table_id)
db_name = connection_manager.get_django_db_alias(config.engine_id)
query = ('SELECT * FROM "%s"' % table_name, [])
return estimate_row_count(query, db_name)
