Description
All services on land-shared-one show periodic NATS disconnect/reconnect cycles (roughly every few minutes based on container logs). Not blocking functionality — messages are delivered and the pipeline works — but it creates log noise and could mask real connection issues.
Affected: nimsforestecommerce, landregistry, hetznertreehouse, and other NATS-connected containers on land-shared-one.
Investigate root cause: could be auth refresh restarting connections, keepalive timeouts, or resource pressure on the embedded NATS server.
Nebula's reasoning: Disconnect/reconnect cycles are abnormal behavior — even though messages are delivered, this is a defect in connection stability, not just a nice-to-have improvement. Priority medium because it's not blocking functionality but the log noise actively degrades observability and could mask real connection failures that need immediate attention.