Planned maintenance: migrate server location
Closed this issue · 2 comments
Is your feature request related to a problem? Please describe.
More convenient access to maintain the RecipeRadar server (singular, currently) during the development team's working week would be convenient at the moment.
Describe the solution you'd like
Physical migration of the server from the current location will be performed, and on reconnection it will serve traffic from a different public IP address.
A maintenance window has been scheduled for today (2023-07-22) for approximately 3pm - 5:30pm UTC.
Since a physical relocation of this kind has not been attempted in recent service history, this operator considers that there is a reasonably high possibility - greater than 50% - of an unexpected outage that exceeds the maintenance time window.
The current operational plan is:
- Take a backup of the backend database.
- Power-off the server.
- Physically relocate the server to the updated (next) location.
- Power-on the server at the updated (next) location.
- Request network configuration checks and changes from our network provider (they have provided a maintenance availability time window for this operation).
- Edit DNS settings to match the updated network configuration.
- Perform any network deconfiguration cleanup that is relevant at the previous (current) location.
Describe alternatives you've considered
- Multiple hosting providers have been considered - a selection was made in the interests of developer productivity.
- The possibility of running the service dual-homed was considered, but this requires further architectural testing.
Additional context
It would be useful to introduce improved bandwidth and electricity usage metrics soon after the server relocation takes place.
Migration was largely successful, and service was restored at approximately 5:30pm UTC on 2023-07-22.
There is some non-critical service degradation currently; the backend services remain disabled, meaning that some recipe website domain iconography is not currently displayed within the application.
Service has been fully restored at the updated location, and cleanup at the previous location is complete; closing this as done.