Aglet

Prioritize a rollback with uncertain serving state

Rollback uncertainty is most urgent when the team may believe a failing version is gone while customers still encounter it. Prioritization should focus on what cannot yet be safely concluded and which evidence will resolve that uncertainty.

Decide where the work belongs

  1. Rank remaining exposure

    List environments, routes, cohorts, workers, and queues that may still serve the original version. Give higher priority to broad or critical paths and to boundaries with no independent version evidence.

  2. Rank recovery confidence

    Compare error, completion, latency, and downstream behavior after the rollback with the original symptom. A quiet interval or lower volume is weak evidence; record why confidence is high or low.

  3. Set a decision checkpoint

    Choose the next review before declaring recovery, expanding traffic, or starting another release. Name the version and behavior records required, and assign the owner responsible for the checkpoint.

What to carry forward

The queue decision should identify exposure uncertainty, current impact, recovery confidence, owner, and checkpoint. Keep it active while a failing version could still serve traffic or behavior evidence is too thin to support closure. Raise priority when the current version, affected exposure, or recovery path is unknown; a visible rollback request with stable verified service may need a different review cadence.

Keep the decision with the work.

Use a Work Item in Aglet to record the problem, the evidence you have, and the next decision. Add an owner and priority, then keep updates in the discussion so the next person can follow the reasoning.

Create an account See the product workflow