Path 07Lesson 7 / 8

Resolve a Needs you interruption

Diagnose why an initiative stopped. Choose continuation, replanning, restart, or a human setup action without losing the decision context.

Practitioner10 minReviewed

Published by How we write

What you will learn

  • Use the initiative and run for their different purposes.
  • Distinguish Resume build, Plan again, Try again, and Start over.
  • Resolve shared inbox work with evidence and the correct permission.

Find the reason for the stop

The fictional equipment service is building an employee-access change. Its initiative shows Needs you. Read the reason on the initiative, then inspect the run if an execution attempt exists.

A failed plan, an interrupted build, and a build that could not start are different conditions. Taiga leaves the initiative at Build with the issue to resolve. The line does not continue through that initiative until someone acts.

Use the run to understand the steps, assumptions, and completed work. Use the initiative to decide what happens next.

Choose the action by its effect

ActionEffect
Resume buildContinues from the first incomplete step and retains committed work
Plan againCreates a replacement plan and preserves prior plan history
Try againRetries a build that could not start
Start overBuilds the current plan as a fresh run on a fresh branch

The interface offers actions that fit the current state. Read the specific explanation before proceeding. Start over does not carry over the stopped run’s work; its prior run remains a record.

For a temporary infrastructure failure, first verify that the cause is resolved. Resume when the plan and retained work remain suitable. For a wrong data boundary, correct the relevant context and plan instead of repeating the same implementation.

If the initiative should no longer run now, move it off the line. Inspect the confirmation when planning or building is active, because the move stops that work.

Resolve setup work separately

Action Required is the shared inbox for reviews and tasks that need a person. It can include credentials, accounts, environment variables, package registry connections, and permissions.

For a missing production secret, identify its required name, provider, environment, and owner. Create or configure it through the approved secret process. Do not paste its value into an initiative description or a learning worksheet.

A setup item describes what to do; marking it done does not create the configuration. Complete and verify the actual task, then mark the item done. Read its timing and the current run state: a missing registry can prevent dependency installation, while another task may block deployment later.

Keep the shared record accurate

The inbox belongs to the team. Resolving an item clears it for everyone with access. Use Dismiss only when the item does not apply. Use Snooze when it should return later.

Members, admins, and owners can resolve items within their permissions. Viewers can read but cannot clear the queue. Confirm that the person assigned to unblock work has the required authority.

Check the result of continuation

After acting, inspect the new state and evidence. Confirm that the intended plan ran and the missing condition was resolved. A repeated failure can indicate an unresolved assumption rather than a need for another identical retry.

Record the enduring lesson in the appropriate instructions or operating procedure. Continue with enabling Taiga across the full responsibility model.

Do the exercise

Classify three fictional interruptions: a temporary package service outage, a plan using the wrong data boundary, and a missing production secret. Name the record to inspect, the owner, the next action, and the evidence that would permit continuation.

Download worksheet (Markdown)

Check your understanding

A build stopped after several completed steps because of a temporary infrastructure failure. The plan remains correct. Which action preserves completed work?

Sources & further reading

Related reading from Taiga