Real-time operational status for all S.O.U.V.E.R.A.I.N. infrastructure components.
Resolved: The elevated latency affecting MQTT wearable data ingestion for users in the EU-West region has been resolved. Root cause: TCP connection pool exhaustion in the MQTT broker under sustained high-frequency HRV stream load. Fix: connection pool size increased 3×. No data was lost.
Resolved: Simulation requests were queuing for 4–8 minutes instead of the typical <90 seconds. Root cause: a memory leak in the generative model serving layer under high concurrency. The affected workers were restarted and the memory issue patched. All queued simulations completed successfully.
Resolved: A subset of causal query requests were returning 503 due to graph store connection timeouts. Affected approximately 3% of API requests for a 22-minute window. No intervention data was affected; the Sovereign Advice queue uses a separate read replica.