What is the main idea of the passage?

A planned rollback can restore stable software quickly, though teams must account for data changes and judge evidence carefully. (The passage explains the safeguard, its limits, and the evidence used to trigger it.)

Why can a database change make rollback difficult?

The older software may not understand information written in a new format. (The third paragraph explains that code can return while changed data remains incompatible.)

What does “thresholds” mean in the final paragraph?

preset levels that trigger attention or action (Signals are monitored against defined levels to decide whether conditions justify action.)

What rollout method limits how many users face a new problem?

a gradual rollout (The final paragraph directly says gradual rollouts limit exposure while evidence accumulates.)

When Software Rolls Back: Level 4 Reading Comprehension Questions

A software team releases a new version of an online service. Minutes later, error reports rise. Customers cannot complete purchases, and the support queue grows. The team could keep repairing the new version while everyone waits. Instead, it may roll back the release.

A rollback restores a previously working version. Modern deployment systems often keep an older package ready, along with records of settings and database changes. Switching back can reduce harm quickly and give engineers time to investigate without leaving every user inside the experiment.

Rolling back is not the same as reversing time. Some actions may already have changed stored data. A new version might write information in a format the old version cannot read. Customers may have submitted orders between the release and the rollback. Teams plan migrations carefully so changes can coexist with both versions, or they prepare a separate recovery step.

The decision also uses evidence. A small rise in harmless warnings may not justify disruption. Failed payments or lost records demand faster action. Teams define signals and thresholds before release, then monitor them. Gradual rollouts limit the number of users exposed while evidence accumulates. A rollback is therefore not an admission that careful engineering failed. It is one of the safeguards careful engineering arranges before uncertainty meets real use.

Comprehension questions