DDLS-23 create weekly integration restore job

[jamesrwarren]

Purpose

Run our restore on integration on a weekly basis to ensure it's fit for purpose in case of disaster.

Fixes DDLS-23

Approach

Checking for a point in time recovery and then a restore from a cross account snapshot.

The whole process is very very slow as that's how aws backups and restores work. The alternative is to pgdump and restore the data but we don't do that for real restores as this doesn't encrypt our data at rest in the backup account and doesn't allow for point in time recovery in the main account.

So anyway the approach here was to create some new roles in org infra and allow our preprod oidc role to assume them. We then run both restores back to back.

Learning

Everything about this takes ages to test so there's a lot of waiting for things to complete.. As such I have managed to test a full run of the cross acct backup to integration from the github pipeline and pit recovery from local but haven't run them both back to back in the pipeline. It ought to work though and we can fix forward if not.

Checklist

• I have performed a self-review of my own code
• I have updated documentation (Confluence/ADR/tech debt doc) where relevant
• I have added tests to prove my work
• The product team have approved these changes
• I have checked my work for potential security issues and refered to the OWASP top 10

Frontend

• I have run an in-browser accessibility test (e.g. WAVE, Lighthouse)
• There are no deprecated CSS classes noted in the profiler
• Translations are used and the profiler doesn't identify any missing
• Any links or buttons added are screen reader friendly and contextually complete
• If adding GA events, I have updated or checked the existing category or label values