Currently, the PITR mechanism on the managed DB is slightly broken in terms of the UI and the API.
On the UI, I'd expect when I select a HHMMSS for point in time recovery, for the DB to be recovered upto that point. But it turns out that this doesn't happen and only the latest backup before that time is used. This is unexpected and a wrong behaviour for the offering.
On the API, there seems to be a limitation of not being able to use uppercase characters in the name field. This returns a blank "422 unprocessable Entity" error without informed where exactly the problem is. It would be good to have more detailed errors in the API instead of having to reach to ask the customer support
The intended behavior is actual PITR; there's a UI bug that is causing it to fallback to the last backup, which we're in the process of fixing now.
I'll take a look at the mentioned API issue.
Attachments Open full size