Aglet

Verify scheduled runs in available and unavailable worker windows

Verification must cover both normal dispatch and the condition that made the run disappear or arrive late. Use the same schedule fixture and read every transition. Confirm the system records an explicit due, delayed, replayed, skipped, or terminal state instead of leaving an operator to infer what happened.

Check whether the outcome improved

  1. Verify normal dispatch

    Let a due occurrence reach enqueue, ready, claim, start, and terminal outcome with workers available. Confirm one occurrence ID is used throughout and the schedule records its result. Preserve this as the handoff control for delayed cases.

  2. Verify delayed availability

    Make workers unavailable or occupied as the occurrence becomes due, then restore service. Check whether the run waits, expires, retries, or enters review according to policy. Confirm due and claim times remain distinct and no second occurrence hides the first.

  3. Verify recovery and readback

    Apply the documented replay or skip decision after inspecting the original occurrence and outputs. Confirm the recovery has a distinct identity and does not duplicate an already applied effect. Read the schedule from a fresh request after both paths settle.

What to carry forward

Accept when available, delayed, and recovery cases produce explicit occurrence identities, handoffs, terminal states, and replay decisions. Keep worker delay separate from missing scheduler emission. Reopen verification if a due run has no durable owner or recovery can create duplicate effects without readback.

Technical background: Rails guides.

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