1.39.10 — Complete infrastructure health evidence

2026-10-23

Complete infrastructure health evidence for existing AWS installations and independent job-runner and scheduler liveness.

Changed

• Fresh host CPU, memory and disk measurements satisfy host health coverage; CloudWatch host metrics remain optional supporting evidence.

Fixed

• Bind managed database and cache metrics to the configured AWS region and exact runtime endpoints, including current Aurora writers and replica-only cache lag.
• Let existing read-only platform monitoring accounts read infrastructure health evidence without granting general cloud inventory access.
• Collect runner and scheduler liveness through existing OS cron, independent of the job queue, and include the original evidence in snapshots.
• Preserve logical node aliases and checked process identity; missing, stale, ambiguous or unproved evidence remains Unknown.

Upgrade notes

• Restart job workers and schedulers after upgrading; retain existing OS cron and expected host/channel declarations.
• Host resource measurements require psutil in normal application deployment dependencies.
• Enrolled sources collect liveness quietly while legacy health notifications remain disabled.