- the bootstrap script needs to be run before the app will run, so mention
that first
- explain what run_celery_beat.sh is for and that it isn't required to run the
app
- seems phonenumber/emailaddress from the CSV are now passed in as personalisation.
- assume the renderer does the correct thing here. Will need to check with @quis
This PR fixes that and adds a test for it.
I am confused as to why I had to change the test_validators test that is checking if the mock is called.
Why did this code pass on preview?
Created a new schema that accepts request parameters for the
get_notifications v2 route.
Using that to validate now instead of the marshmallow validation.
Also changed the way formatted error messages are returned because
the previous way was cutting off our failing `enum` messages.
Switched OFF on DEV and by DEFAULT
Switched ON for TESTS
Switched ON on PREVIEW - STAGING - PRODUCTION
URL set via ENV file - set in environment_test.sh. DEVS will need to update environment.sh. README updated.
In this PR the id for the notification is passed in and used to created the notification, which causes a integrity error.
Normally when we get a SQLAlchemy error here we send the message to the retry queue, but if the notification already exists
we just ignore it.
There are no more notifications whose statuses are "failed", as
the "failed" status has now been replaced with statuses that are
more specific about the nature of the failure.
However, we still want to be able to filter by failing
notifications. (ie "/v2/notifications?status=failed").
Created a `.substitute_status()` method which takes a status
string or list of status strings and, if it finds 'failure',
replaces it with the other failing status types.
This way, calling for nottifications with "?status=failed" is
internally treated as
"status = ['technical-failure', 'temporary-failure', 'permanent-failure']"
Some notification statuses assume that a notification has been
updated (ie, it cannot have been created in that state).
This caused a bug in our sample notification fixture when trying
to create a notificaiton in a 'complete' status.
This commit groups the completed statuses in a list, the way other
statuses have been grouped together so that they're more portable.
Also fixed the sample_notification fixture.
Return multiple notifications for a service.
Choosing a page_size or a page_number is no longer allowed.
Instead, there is a `next` link included with will return the
next {default_page_size} notifications in the sequence.
Query parameters accepted are:
- template_type: filter by specific template types
- status: filter by specific statuses
- older_than: return a chronological list of notifications older
than this one. The notification with the id that is passed in
is _not_ returned.
Note that both `template_type` and `status` can accept multiple
parameters. Thus it is possible to call
`/v2/notifications?status=created&status=sending&status=delivered`
The "cost" value was flawed for a couple of reasons.
1. Lots of messages are free, so in those instances the "cost"
doesn't tell you anything
2. The query to get the rate was expensive and we don't have
an obvious way to get it back very efficiently for large numbers
of notifications.
So we scrapped it.
We want to log the usage of the various API clients we have so that we understand when they can be cycled.
To this end we are going to count usage in statsd.
All notify clients have a suer agent, of the format: NOTIFY-API-{LANGUAGE}-CLIENT/version.number
For example, NOTIFY-API-PYTHON-CLIENT/3.0.0
We convert that into a statsd/graphite friendly string of the format: notify-api-python-client.3-0-0
So we can subdivide on client and client version on our dashboards.
Present but unknown User agents are records as "non-notify-user-agent"
Missing are presented as "unknown"
