Lockesmith on Nostr: It's got nothing to do with developers and everything to do with the people ...

It's got nothing to do with developers and everything to do with the people administering the systems.

There is little to no excuse for designing a system in which you cannot spin up the upgraded version in parallel, make sure everything is working, and cut to it.

This is known as a blue-green deployment.
A full upgrade, other than a kernel level upgrade should never require a full shutdown of more than a few minutes. And if proper redundancy is in place, that would be a series of rolling restarts, in which case the end user would never even notice.

Neither snapshots nor backups should cause a full system outage. One server at a time, maybe. But end users should never feel the effect of that.

It doesn't matter whether the service is legacy or not, anyone taking people's money for a service should have implemented proper failover and redundancy. We live in a time where there are no valid excuses for this behavior.

Fixing this has nothing to do with any freedom tech, only integrity of the companies and individuals running the individual systems.