bump staging RDS instance to T3 Small

[freemvmt]

Related to this ticket.

This PR bumps the RDS PostgreSQL DB instance used on staging to db.t3.small, in order to increase the number of connections it can handle by doubling the memory (to 2GB).

Prod is still pinned to db.t3.medium, which has twice the memory again (4GB).

We only need to merge this if #4080 doesn't solve the scaling issue - also see that PR for further context and motivation for this change.

Also note that we'll have to manually roll out the upgrade on AWS via Pulumi CLI, since there is no CI setup for the data layer as there is for application (see the push-main workflow).

[github-actions]

Removed vultr server and associated DNS entries

[DafyddLlyr]

Approved so that you can merge this should you wish to do so 👍

Also note that we'll have to manually roll out the upgrade on AWS via Pulumi CLI, since there is no CI setup for the data layer as there is for application (see the push-main workflow).

Very well remembered! We can run pulumi up --stack staging from the /infrastructure/data folder locally in order to achieve this - there's an intermediate step where you'll get told if there are any destructive action (the classic IaC +, -, ~).

I'd expect this to show a change ~ to instance class which would be non-destructive (i.e. not a + / -). I think that the change will be auto-applied at the next maintenance window but I'm not totally sure, the might actually be a pulumi flag to force the change.

If the absolute worst happens (very very unlikely) there's nothing irreplaceable on the staging db - it's just a clone of a subset of data from production which we can re-populate by running the "Nightly staging database sync" GitHub action.

[freemvmt]

@DafyddLlyr amazing thanks for all that additional context, especially re the appropriate action to run for recovery from everything going sideways ;)

Next maintenance window isn't till next Wednesday so I probably would look to force it, so thanks on that point too!

[freemvmt]

@DafyddLlyr right you were!

However I apparently didn't have the right permissions to make this change from local, so I actually just made the change directly in the AWS console as we discussed the other day. Will merge this PR to reflect that the change has been applied.