- Replace "custom protocol" placeholder with full LmxProxy details (gRPC transport, SDK API mapping, session management, keep-alive, TLS, batch ops)
- Add bullet-level requirement traceability, design constraint traceability (52 KDD + 6 CD), split-section tracking, and post-generation orphan check to plan framework
- Resolve Q9 (LmxProxy), Q11 (REST test server), Q13 (solo dev), Q14 (self-test), Q15 (Machine Data DB out of scope)
- Set Central UI constraints: Blazor Server + Bootstrap only, no heavy frameworks, custom components, clean corporate design
Cluster Infrastructure: add min-nr-of-members=1 requirement for single-node
operation after failover. Add graceful shutdown / CoordinatedShutdown section
for fast singleton handover during planned maintenance.
Site Runtime: add explicit supervision strategies per actor type (Resume for
coordinators, Stop for short-lived execution actors). Stagger Instance Actor
startup to prevent reconnection storms. Add Tell-vs-Ask usage guidance per
Akka.NET best practices (Tell for hot path, Ask for system boundaries only).
Data Connection Layer: add Connection Actor Model section documenting the
Become/Stash pattern for connection lifecycle state machine.
Health Monitoring: add dead letter count as a monitored metric.
Host: add REQ-HOST-8a for dead letter monitoring (subscribe to EventStream,
log at Warning level, report as health metric).
Add keep-oldest split-brain resolver with 15s stable-after duration. Configure both
nodes as seed nodes for symmetric startup. Set moderate failure detection defaults
(2s heartbeat, 10s threshold, ~25s total failover). Document automatic dual-node
recovery from persistent storage with no manual intervention.
