What happens if an update fails?

We are prepared for an update failure rather than improvising. Before every update, we create a complete, tested backup and verify the update on staging beforehand. Should an update cause problems on the live system despite these tests, we perform a rollback to the last stable state within minutes.

In parallel, post-update monitoring detects deviations early, often before end customers notice anything, by watching critical metrics such as response times and error rates right after deployment. This combination of tested backup, fast rollback and close observation is the reason why controlled security updates are considerably less risky than untested updates applied alone.