ServiceRadar v1.0.92

Performance and reliability improvements for the observability dashboard and RBAC system.

Whats New

1.0.92

• RBAC permission lookups now use a tiered cache: in-memory ETS first, then database failover — eliminating redundant DB round-trips on every request.
• Observability page queries are dramatically faster: replaced coalesce(observed_timestamp, timestamp) with direct timestamp filtering, enabling TimescaleDB chunk exclusion and index usage instead of full sequential scans.
• Added NATS ingest notifier for live dashboard refresh: a lightweight GenServer subscribes to processed log/flow subjects and bridges notifications to PubSub with debounce, so the observability page updates automatically when new data arrives.
• Fixed device creation via the UI by removing the pre-read race and handling unique UID conflicts cleanly.
• Fixed device delete and restore actions (proper deleted_by/reason handling plus stale-record recovery).
• Added error logging for device create/delete/restore failures so issues show up in server logs.
• Updated the netflow collector to parser 0.9.0 with cached flows, capped pending flows, and refreshed test fixtures.