Gossip, push/pull, and the missing window: multi-region Alertmanager dedup
How a push/pull interval larger than the peer timeout broke deduplication between two Alertmanagers in different regions, and what I learned validating HA under failure.
Home / Field notes
Lessons from
building and operating.
Investigations, architecture decisions, and practical notes from working with real systems. Written to make the reasoning useful to someone else.
Implementing RBAC: when centralizing looked like the right answer
Why a centralized RBAC Core turned into a single point of failure and an organizational bottleneck, and how a federated model with OpenFGA fixed it.
Docker → Kubernetes (K3s) in a Multi-Tenant Environment
A practical checklist from the K3s migration: rollout, per-tenant isolation, and the trade-offs that kept incidents at bay.