This PR will optimize this query to use a more efficient index.
- Add notification_type to the dao_get_last_template_usage to optimize the query.
- Tested and analyzed query on production database with very significant results.
Before:
QUERY PLAN
-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Limit (cost=0.43..1711.35 rows=1 width=935) (actual time=21186.053..21186.053 rows=0 loops=1)
-> Index Scan Backward using ix_notifications_created_at on notifications (cost=0.43..4607493.80 rows=2693 width=935) (actual time=21186.052..21186.052 rows=0 loops=1)
Filter: (((key_type)::text <> 'test'::text) AND (template_id = 'xxxxxx'::uuid))
Rows Removed by Filter: 8244071
Planning time: 0.112 ms
Execution time: 21186.082 ms
After:
QUERY PLAN
---------------------------------------------------------------------------------------------------------------------
Limit (cost=5323.10..5323.10 rows=1 width=935)
-> Sort (cost=5323.10..5323.74 rows=258 width=935)
Sort Key: created_at DESC
-> Index Scan using ix_notifications_template_id on notifications (cost=0.56..5321.81 rows=258 width=935)
Index Cond: (template_id = 'xxxxx'::uuid)
Filter: (((key_type)::text <> 'test'::text) AND (notification_type = 'sms'::notification_type))
Planning time: 1.102 ms
Execution time: 0.584 ms
We need to deal with this, it's ok when updating a notification status from delivered to delivered. But the DailySortedLetter counts are being doubled.
Adding the file_name to the table as a unique key to get around this issue. It will mean we have multiple rows for each billing_day, but that's ok we can aggregate that.
This will also give us a way to see which file created which count.
SQLAlchemy handles escaping anything that could allow a SQL injection
attack. But it doesn’t escape the characters used for wildcard
searching. This is the reason we’re able to do `.like('%example%')`
at all.
But we shouldn’t be letting our users search with wildcard characters,
so we need to escape them. Which is what this commit does.
* Deleted the statistics DAO
(this was used for the job statistics tasks)
* Deleted the functions in the jobs DAO which are no longer used
(the functions that were used for the job-stats endpoints)
It is possible to search for a phone number when from the email notification page and get a SMS message in return.
This also helps to optimise the query.
Phone numbers sometimes contain stuff we normalise out. This matches
perfectly if we have a full phone number, because we can normalise the
thing we’re searching for in the same way as the search term.
With partial search terms we can’t do this completely, because we can’t
work out if ‘123’ is part of a UK number, an international number, the
start of the phone number, the last 3 digits, etc.
What we can do is remove some stuff that we can know will cause partial
search terms to not match:
- leading pluses
- leading `0`s
- any brackets
- any spaces
Users expect the search to work on partial email addresses ‘similar to
Gov.Pay’. We can’t have a situation where Pay are doing something better
than us 😜
This will continue to update the notification history for letter notifications.
We currently have an issue where the responses to letters from the provider is taking a long time.
This is due to the manual nature of their process.
Updating the status of the letter will still work if the notification has been purged.
Also turned back on the purge letter notification scheduled task.
Filtering out hidden templates requires all existing templates to
have `hidden` flag set, which can only be done by a migration after
the code that sets the flag to `False` by default for new templates
has been released.
This removes the filtering logic until the migration has been released.
Switched to using a isinstance check on the string.
Added an order by clause to dao_get_template_usage_stats_by_service, it was causing an itermitten failure in the tests.
- Changed the organisation DAO update method to only make 1 query
- Updated the update rest endpoint to not return an organisation when
the update is successful
they were not included in nightly task since that runs off
NotificationHistory, which doesn't include test keys. However, when you
load the page we top up the nightly stats with today's data from the
Notifications table, which *does* include test data.
notable things that have been kept until migration is complete:
* passing in `organisation` to update_service will update email branding
* both `/email-branding` and `/organisation` hit the same code
* service endpoints still return organisation as well as email branding
if the international_billing_rates.yml has `dlr: null`, that means we
don't know what delivery receipts they provide - they might not provide
any. So if we do get an update, we don't know for sure that the message
was actually delivered - lets not update it.
Letters is a mature enough feature now – and one that we’ve been talking
about offering for long enough – that we shouldn’t make people dig
around in the settings.
I think we’d want to wait a bit longer/indefinitely before deciding to
turn it on for existing services across the platform.
The history was not being updated properly, we think this is because the declaritive attribute is not being set propery by the property.
When reply_to: None it will update the service_letter_contact_id, but not the service_letter_contact, we think when the history_meta is build the history class and checking if the value is updated it depends which attribute it is checking first.
In order to fix this issue, there is a new dao method to update the reply_to on the Template and insert a new Template history.
