# Safe Scheduled Runtime Upgrade

- id: safe-scheduled-runtime-upgrade
- type: skill
- category: production-reliability
- detail: https://agents.proofandstate.com/products/safe-scheduled-runtime-upgrade

## Summary
Upgrade scheduled runtimes through an exact target, verified job draining, a fresh guarded review, recoverable timer restoration, and separate process-readiness and operational-completion checks.

## Problem
Scheduled workers and queued jobs can change state during a release, while interrupted switches, automatic catch-up work, and slow diagnostics can make repeated upgrades or premature success claims unsafe.

## Core outcomes
- Exact target and clean-source verification with recovery ownership
- Runtime, trigger, service-dependency and queued-job inventory
- Bounded draining with preservation of original scheduling states
- Fresh state-bound review and a single guarded release application
- Recovery decisions for stale reviews, cancellation and ambiguous switches
- Process-reported runtime readiness separated from pending asynchronous work
- Configuration restart planning with full stop scope and durable handoff

## Requirements
- A runtime driven by systemd timers or a scheduler with equivalent inspection and pause controls
- An exact target revision or artifact digest and reviewable source
- Access to service, process, queued-job and release-state observations
- A release mechanism that rejects stale state or equivalent concurrent drift
- Authority for the affected operational changes and a named recovery owner
- Durable recovery evidence and a readiness check identifying the running instance

## Boundaries
- A source checkout or successful apply response does not establish running-process readiness
- Paused timers and empty PID fields do not establish that queued or in-flight work has drained
- Worker termination, lock deletion and receipt rewriting are not shortcuts to quiescence
- An ambiguous switch must be reconciled before repetition or rollback
- Resuming schedules may start consequential catch-up work and requires the agreed recovery rule
- Asynchronous start acceptance and slow diagnostic snapshots do not prove operational completion
- Host and virtual-machine restarts require authority covering their full stop scope
- The workflow provides no universal safety guarantee or claim of live acceptance

## Tags
- runtime-upgrade
- scheduled-workers
- systemd
- job-draining
- state-bound-review
- recovery
- process-readiness
- wsl

## Canonical source
- repository: AyobamiH/agent-shop-products (main)
- payload path: products/safe-scheduled-runtime-upgrade/SKILL.md

## Not published
Price, ratings, reviews, sales figures and full PROMPT.md / SKILL.md payload bodies are not published.