We've already performed this update on production since you need to create the index concurrently, which is not allowed from the alembic script. For that reason we are checking if the index exists.
We are adding an index to Notifications to optimize the get_notifications_for_service. We need to build the index concurrently which can not be run inside a transaction block so the index will need to be run on the db directly.
CREATE INDEX CONCURRENTLY ix_notifications_service_created_at ON notifications (service_id, created_at);
DROP INDEX CONCURRENTLY ix_notifications_service_created_at
We don't use FUNCTIONAL_TEST_PROVIDER_SERVICE_ID or
UNCTIONAL_TEST_PROVIDER_SMS_TEMPLATE_ID anymore so we can safely
delete them from config and tests.
sent_by_email_address field was added because sometimes two
people at one institution have the same name and then email
address, which is unique, is more useful.
The delivery for provider is slow if more than threshold (currently
we pass in threshold 10%) either took x (for now 4) minutes to deliver,
or are still sending after that time. We look at all notifications
for current provider which are delivered or sending, and are not under
test key, for the last 10 minutes.
We are using created_at to establish if notifications are from last
10 minutes because we have an index on it, so the query is faster.
Also write tests for new is_delivery_slow_for_provider query
We don't seem to use recorded queries or modification tracking anywhere
in the app, and both features potentially increase memory usage.
This removes deprecated SQLALCHEMY_COMMIT_ON_TEARDOWN options. It's
been removed from the docs and the default matches the value we set
anyway.
Added cancelled letters to the number of failed letters in the statistics
that get used for the dashboard. At some point, we want to stop
including cancelled letters in the stats, but for now this keeps things
consistent with our current letter failure state, permanent-failure.
It can be useful to get a notification by id while checking that the
notification belongs to a given service. This changes the
get_notification_by_id DAO function to optionally also filter by
service_id so that we can check this.
Letters should always have a reference, because that’s what DVLA use to
tell us when they’ve sent a letter.
If a letter has a reference of `None` then DVLA say they’ve sent a
letter with a reference of `'None'`. This means we can never reconcile
the letter, which means it stays in `created`, which means it never
gets billed.
We don’t think this has affected any real letters yet, just ones that
we’ve sent as tests.
By default, SQLAlchemy will start a transaction with an
existing connection without checking that the connection
is still valid.
Enabling "pre-ping" makes the ORM send a `SELECT 1` when
acquiring a connection, which should help avoid some errors
caused by connections breaking during a DB failover.
The added statement has a constant overhead for all transactions,
so we should only keep it enabled when we're planning to switch
or upgrade the database server.
https://docs.sqlalchemy.org/en/latest/core/pooling.html#disconnect-handling-pessimistic
Running an update before setting the column default value reduces
the time the table is locked (since most rows don't have a NULL
value anymore), but the migration takes slightly longer to run
overall.
Also test deleting jobs with flexible data retention
Also update tests for default data retention following logic
change: dao_get_jobs_older_than_data_retention now counts
today at the start of the day, not at a time when function runs
and updated tests reflect that
Services reaching rate limits are triggering our alerts and make it
hard to find actual exceptions in the logs.
As far as the API is concerned this is not an exceptional state,
so we shouldn't log it as such.
