[Infrastructure] Graceful shutdown, Redis snapshot config, DaemonSet update strategy

[Flegma]

Summary

Operational configuration gaps affecting service reliability.

Findings

• All deployments — no terminationGracePeriodSeconds or preStop hooks.
• Redis configmap — save disables RDB snapshots. noeviction errors when full.
• Connector DaemonSet — no explicit maxUnavailable. All nodes could update simultaneously.

Impact

Abrupt termination drops requests. Redis data loss on crash. All game nodes offline during update.

Suggested Fix

Add terminationGracePeriodSeconds: 60. Enable RDB snapshots. Set maxUnavailable: 1 on DaemonSet.