Confirming and running the reassignment

This covers steps 5–6: the binding permission check at confirm time, and running the reassignment safely.

Step 5 — confirm (a binding permission check)

When you move from Preview to Confirm, the app re-checks permissions before any write happens:

• it re-validates your Administer Jira permission, and
• it checks your access per project for the projects involved.

If a project fails the check, the run is blocked with a reason naming that project — the reassignment is not partially applied behind a green check. This is a binding gate, not an advisory note: the app would rather stop and tell you than write a half-finished result.

The app also re-validates the target is still active and captures a point-in-time snapshot at confirm, so that if your authority changes mid-run the run pauses rather than continuing on stale permissions (see Resuming below).

When the check passes, you confirm, and the run begins.

Step 6 — run (and resume if interrupted)

The reassignment runs in the background and reports progress per content type. Two things keep it safe:

• It's safe to retry. Each write is performed so that re-running it doesn't double-apply — if the run is interrupted, it resumes from where it left off and re-reads the final state rather than blindly repeating work.
• Partial failures are visible. If an individual item can't be reassigned, that item is shown as failed with its reason — it is never hidden behind an overall green result. The run finishes as completed or completed with failures, and the failed items are surfaced plainly.

Resuming a paused run

A run can pause itself — for example, if the per-project permission picture changes during the run (a time-of-check/time-of-use safeguard). When that happens you'll see a paused / resumable state with the reason. Resolve the cause (for example, restore the access that changed) and select Resume; the run continues and rebuilds its counts correctly rather than starting over or double-counting.

After the run

Once the run reaches a terminal state, move to the closure-evidence export and retention. If some items failed, the evidence makes that explicit so your record reflects what actually happened.