Merge pull request #3454 from alphagov/upsert-status-180693991

Rewrite status aggregation to be a bulk upsert
This commit is contained in:
Ben Thorner
2022-02-17 13:21:50 +00:00
committed by GitHub
4 changed files with 72 additions and 69 deletions

View File

@@ -10,10 +10,7 @@ from app.dao.fact_billing_dao import (
fetch_billing_data_for_day, fetch_billing_data_for_day,
update_fact_billing, update_fact_billing,
) )
from app.dao.fact_notification_status_dao import ( from app.dao.fact_notification_status_dao import update_fact_notification_status
fetch_status_data_for_service_and_day,
update_fact_notification_status,
)
from app.dao.notifications_dao import get_service_ids_with_notifications_on_date from app.dao.notifications_dao import get_service_ids_with_notifications_on_date
from app.models import EMAIL_TYPE, LETTER_TYPE, SMS_TYPE from app.models import EMAIL_TYPE, LETTER_TYPE, SMS_TYPE
@@ -117,23 +114,8 @@ def create_nightly_notification_status():
def create_nightly_notification_status_for_service_and_day(process_day, service_id, notification_type): def create_nightly_notification_status_for_service_and_day(process_day, service_id, notification_type):
process_day = datetime.strptime(process_day, "%Y-%m-%d").date() process_day = datetime.strptime(process_day, "%Y-%m-%d").date()
start = datetime.utcnow()
new_status_rows = fetch_status_data_for_service_and_day(
process_day=process_day,
notification_type=notification_type,
service_id=service_id,
)
end = datetime.utcnow()
current_app.logger.info(
f'create-nightly-notification-status-for-service-and-day task fetch '
f'for {service_id}, {notification_type} for {process_day}: '
f'data fetched in {(end - start).seconds} seconds'
)
start = datetime.utcnow() start = datetime.utcnow()
update_fact_notification_status( update_fact_notification_status(
new_status_rows=new_status_rows,
process_day=process_day, process_day=process_day,
notification_type=notification_type, notification_type=notification_type,
service_id=service_id service_id=service_id
@@ -143,5 +125,5 @@ def create_nightly_notification_status_for_service_and_day(process_day, service_
current_app.logger.info( current_app.logger.info(
f'create-nightly-notification-status-for-service-and-day task update ' f'create-nightly-notification-status-for-service-and-day task update '
f'for {service_id}, {notification_type} for {process_day}: ' f'for {service_id}, {notification_type} for {process_day}: '
f'data updated in {(end - start).seconds} seconds' f'updated in {(end - start).seconds} seconds'
) )

View File

@@ -34,56 +34,65 @@ from app.utils import (
) )
def fetch_status_data_for_service_and_day(process_day, service_id, notification_type): @autocommit
def update_fact_notification_status(process_day, notification_type, service_id):
start_date = get_london_midnight_in_utc(process_day) start_date = get_london_midnight_in_utc(process_day)
end_date = get_london_midnight_in_utc(process_day + timedelta(days=1)) end_date = get_london_midnight_in_utc(process_day + timedelta(days=1))
# query notifications or notification_history for the day, depending on their data retention # delete any existing rows in case some no longer exist e.g. if all messages are sent
service = Service.query.get(service_id)
table = get_notification_table_to_use(service, notification_type, process_day, has_delete_task_run=False)
return db.session.query(
table.template_id,
func.coalesce(table.job_id, '00000000-0000-0000-0000-000000000000').label('job_id'),
table.key_type,
table.status,
func.count().label('notification_count')
).filter(
table.created_at >= start_date,
table.created_at < end_date,
table.notification_type == notification_type,
table.service_id == service_id,
table.key_type.in_((KEY_TYPE_NORMAL, KEY_TYPE_TEAM)),
).group_by(
table.template_id,
'job_id',
table.key_type,
table.status
).all()
@autocommit
def update_fact_notification_status(new_status_rows, process_day, notification_type, service_id):
table = FactNotificationStatus.__table__
FactNotificationStatus.query.filter( FactNotificationStatus.query.filter(
FactNotificationStatus.bst_date == process_day, FactNotificationStatus.bst_date == process_day,
FactNotificationStatus.notification_type == notification_type, FactNotificationStatus.notification_type == notification_type,
FactNotificationStatus.service_id == service_id, FactNotificationStatus.service_id == service_id,
).delete() ).delete()
for row in new_status_rows: # query notifications or notification_history for the day, depending on their data retention
db.session.connection().execute( service = Service.query.get(service_id)
insert(table).values( source_table = get_notification_table_to_use(
bst_date=process_day, service,
template_id=row.template_id, notification_type,
service_id=service_id, process_day,
job_id=row.job_id, has_delete_task_run=False
notification_type=notification_type, )
key_type=row.key_type,
notification_status=row.status, query = db.session.query(
notification_count=row.notification_count, literal(process_day).label("process_day"),
) source_table.template_id,
literal(service_id).label("service_id"),
func.coalesce(source_table.job_id, '00000000-0000-0000-0000-000000000000').label('job_id'),
literal(notification_type).label("notification_type"),
source_table.key_type,
source_table.status,
func.count().label('notification_count')
).filter(
source_table.created_at >= start_date,
source_table.created_at < end_date,
source_table.notification_type == notification_type,
source_table.service_id == service_id,
source_table.key_type.in_((KEY_TYPE_NORMAL, KEY_TYPE_TEAM)),
).group_by(
source_table.template_id,
source_table.template_id,
'job_id',
source_table.key_type,
source_table.status
)
db.session.connection().execute(
insert(FactNotificationStatus.__table__).from_select(
[
FactNotificationStatus.bst_date,
FactNotificationStatus.template_id,
FactNotificationStatus.service_id,
FactNotificationStatus.job_id,
FactNotificationStatus.notification_type,
FactNotificationStatus.key_type,
FactNotificationStatus.notification_status,
FactNotificationStatus.notification_count
],
query
) )
)
def fetch_notification_status_for_service_by_month(start_date, end_date, service_id): def fetch_notification_status_for_service_by_month(start_date, end_date, service_id):

View File

@@ -599,9 +599,10 @@ def test_create_nightly_notification_status_for_service_and_day(notify_db_sessio
def test_create_nightly_notification_status_for_service_and_day_overwrites_old_data(notify_db_session): def test_create_nightly_notification_status_for_service_and_day_overwrites_old_data(notify_db_session):
first_service = create_service(service_name='First Service') first_service = create_service(service_name='First Service')
first_template = create_template(service=first_service) first_template = create_template(service=first_service)
create_notification(template=first_template, status='delivered')
process_day = date.today() process_day = date.today()
# first run: one notification, expect one row (just one status)
notification = create_notification(template=first_template, status='sending')
create_nightly_notification_status_for_service_and_day(str(process_day), first_service.id, 'sms') create_nightly_notification_status_for_service_and_day(str(process_day), first_service.id, 'sms')
new_fact_data = FactNotificationStatus.query.order_by( new_fact_data = FactNotificationStatus.query.order_by(
@@ -611,8 +612,11 @@ def test_create_nightly_notification_status_for_service_and_day_overwrites_old_d
assert len(new_fact_data) == 1 assert len(new_fact_data) == 1
assert new_fact_data[0].notification_count == 1 assert new_fact_data[0].notification_count == 1
assert new_fact_data[0].notification_status == 'sending'
create_notification(template=first_template, status='delivered') # second run: status changed, still expect one row (one status)
notification.status = 'delivered'
create_notification(template=first_template, status='created')
create_nightly_notification_status_for_service_and_day(str(process_day), first_service.id, 'sms') create_nightly_notification_status_for_service_and_day(str(process_day), first_service.id, 'sms')
updated_fact_data = FactNotificationStatus.query.order_by( updated_fact_data = FactNotificationStatus.query.order_by(
@@ -620,8 +624,11 @@ def test_create_nightly_notification_status_for_service_and_day_overwrites_old_d
FactNotificationStatus.notification_type FactNotificationStatus.notification_type
).all() ).all()
assert len(updated_fact_data) == 1 assert len(updated_fact_data) == 2
assert updated_fact_data[0].notification_count == 2 assert updated_fact_data[0].notification_count == 1
assert updated_fact_data[0].notification_status == 'created'
assert updated_fact_data[1].notification_count == 1
assert updated_fact_data[1].notification_status == 'delivered'
# the job runs at 12:30am London time. 04/01 is in BST. # the job runs at 12:30am London time. 04/01 is in BST.

View File

@@ -14,8 +14,8 @@ from app.dao.fact_notification_status_dao import (
fetch_notification_status_totals_for_all_services, fetch_notification_status_totals_for_all_services,
fetch_notification_statuses_for_job, fetch_notification_statuses_for_job,
fetch_stats_for_all_services_by_date_range, fetch_stats_for_all_services_by_date_range,
fetch_status_data_for_service_and_day,
get_total_notifications_for_date_range, get_total_notifications_for_date_range,
update_fact_notification_status,
) )
from app.models import ( from app.models import (
EMAIL_TYPE, EMAIL_TYPE,
@@ -32,6 +32,7 @@ from app.models import (
NOTIFICATION_TECHNICAL_FAILURE, NOTIFICATION_TECHNICAL_FAILURE,
NOTIFICATION_TEMPORARY_FAILURE, NOTIFICATION_TEMPORARY_FAILURE,
SMS_TYPE, SMS_TYPE,
FactNotificationStatus,
) )
from tests.app.db import ( from tests.app.db import (
create_ft_notification_status, create_ft_notification_status,
@@ -618,7 +619,7 @@ def test_get_total_notifications_for_date_range(sample_service):
('2022-03-27T23:30', date(2022, 3, 27), 0), # 28/03 00:30 BST ('2022-03-27T23:30', date(2022, 3, 27), 0), # 28/03 00:30 BST
('2022-03-26T23:30', date(2022, 3, 26), 1), # 26/03 23:30 GMT ('2022-03-26T23:30', date(2022, 3, 26), 1), # 26/03 23:30 GMT
]) ])
def test_fetch_status_data_for_service_and_day_respects_gmt_bst( def test_update_fact_notification_status_respects_gmt_bst(
sample_template, sample_template,
sample_service, sample_service,
created_at_utc, created_at_utc,
@@ -626,5 +627,9 @@ def test_fetch_status_data_for_service_and_day_respects_gmt_bst(
expected_count, expected_count,
): ):
create_notification(template=sample_template, created_at=created_at_utc) create_notification(template=sample_template, created_at=created_at_utc)
rows = fetch_status_data_for_service_and_day(process_day, sample_service.id, SMS_TYPE) update_fact_notification_status(process_day, SMS_TYPE, sample_service.id)
assert len(rows) == expected_count
assert FactNotificationStatus.query.filter_by(
service_id=sample_service.id,
bst_date=process_day
).count() == expected_count