Previous:
assumed discrepancy in stats counts to be related to timeouts
Now:
If discrepancy exists do the math on the notifications for that job to work out counts based on statuses to redo stats.
After three days we timeout the notifications that we have not received a receipt for.
In the same way we bump the failed count to match the job count if there is a descepenecy. We do this after the same period that we do for notifications.
jsonschema states:
> A format attribute can generally only validate a given set of
> instance types. If the type of the instance to validate is not in
> this set, validation for this format attribute and instance SHOULD
> succeed.
We were not checking for the type of the input, and our validators were
behaving in unexpected manners (throwing TypeErrors etc etc). Despite
us declaring that the phone_number field is of type `str`, we still
need to make sure the validator passes gracefully, so that the inbuilt
type check can be the bit that catches if someone passes in a non-str
value. We've seen this with people passing in integers instead of strs
for phone numbers. This'll make them receive a nice 400 error
(e.g. "phone_number 12345 is not of type string"), rather than us
having a 500 internal server error
- Looks now for a specific Integrity error, not a generic SQLAlchemy error when checking for an insert failure
- If the update fails post an insert error, then there is an issue so raise an exception so tasks can be retried.
- create_or_update_job_sending_statistics
This will try and update an existing row. if this fails as it hasn't been created, then it will insert a row. If this fails as another process has got there first then it should try and update again.
This is a code version of upset
- update_job_stats_outcome_count
Will update the outcome states. Uses the NOTIFICATION_STATUS_TYPES_FAILED to determine if the notification failed.
Any thing in DELIVERED will be marked as delivered. Statues not in the FAILED array or delivered provoke no update to the table.
We can then use this later - saving an extra DB query on every client facing API call
- Note this doesn't affect admin calls which do not use the service from the api key, but use the one passed as part of the URL path.
We now have a new column in the database, but it isn't being
populated. The first step is to make sure we update this column,
while still keeping the old enum based column up to date as well.
A couple of changes have had to happen to support this - one irritating
thing is that if we're ever querying columns individually, including
`Notification.status`, then we'll need to give that column a label,
since under the hood it translates to `Notification._status_enum`.
Accessing status through the ORM (i.e., my_noti.status = 'sending' or
similar) will work fine.
