Errors and reliability

Required release loop

• Learning-data backup: A release backup exists, is stored outside the web process and has object-count evidence.
• Restore drill: The latest backup can be deserialized and counted without overwriting live learner data.
• Learner-facing smoke check: Core learner, curriculum, safety, readiness and operations routes pass against the intended release build.
• Adaptive-learning load check: Synthetic learner traffic stays within the configured response-time threshold.
• Failure-mode harness: Invalid routes, unreviewed teaching, blocked concepts, safety input and shallow answers fail safely.
• Production preflight: Target deployment settings pass security, host, HTTPS, cookie, HSTS, database and expert-review-gate checks.
• Full operations rehearsal: Backup, restore, smoke, load and failure-mode steps pass in one release-context rehearsal.
• No unresolved production-impacting events: Every open error or critical operational event is resolved or explicitly risk-accepted before sign-off.

Production acceptance rules

• Internal operational evidence: All backup, restore, smoke, load, failure-mode, preflight, rehearsal and open-error checks are passing.
• Target environment sign-off: Environment, HTTPS production URL, monitoring, alerting, backup storage and incident owner are configured for the named release target.
• Production owner acceptance: A production operations owner accepts the evidence in the validation registry with no open handoff items.
• External or real-world attestation: Reviewer confirms the evidence is target-environment or real-world evidence and not synthetic-only.

Next actions

• Run a learning-data backup and keep the generated backup path with the release evidence.
• Verify the latest backup can be deserialized without writing over live learner data.
• Run the learner-facing route smoke check against the intended release build.
• Run the adaptive-learning load check and confirm response timings stay within the release threshold.
• Set ARISTOTLE_DEPLOYMENT_ENVIRONMENT to the named target environment this evidence applies to.
• Set ARISTOTLE_PRODUCTION_URL to the HTTPS learner-facing URL and rerun smoke/preflight checks.
• Add ARISTOTLE_MONITORING_DASHBOARD_URL for the external uptime, error-rate and latency dashboard.
• Add ARISTOTLE_ALERTING_ROUTE_REFERENCE for the human alert route used during learner-impacting incidents.

Return protocol

• Inspect internal operational evidence: Confirm backup, restore, smoke, load, failure-mode, production preflight, operations rehearsal and open-error evidence are current for the release.
• Inspect target-environment sign-off: Confirm the named deployment, HTTPS learner URL, monitoring, alerting, backup storage and incident/runbook owner exist outside the local app.
• Inspect the release loop: Check that operational checks were run in the intended release context and rerun them after deployment, database or curriculum-trust changes.
• Return a production-owner decision: Return accept, changes_requested or reject with reviewer identity, role, attestation reference, review notes and any required release changes.
• Accept validation evidence only after review: A production operations validation record counts only when the owner accepts it with target-environment or real-world attestation and no open handoff items.

Trust boundary

This work order coordinates production-owner review. It does not itself prove deployment readiness, external monitoring, backup retrieval or incident ownership; those claims require accepted validation evidence from the target release context.

The work order includes readiness counts, command names, route references, open-item keys and validation draft summaries only. It omits learner answers, learner identity, monitoring URLs, alerting references, backup storage references and production URLs.