Topics

Posts grouped by area: Kubernetes, observability, networking, storage, databases and the incidents behind them.

Kubernetes

Running Kubernetes in production: upgrades, operators, StatefulSets, probes and the failures that only show up on real clusters.

Metrics, alerts and SLOs that point at the cause: Prometheus, VictoriaMetrics, Alertmanager, Grafana and the signals that lie.

Real incidents taken apart: the first wrong guess, how the root cause was proven and what changed afterwards.

Cluster networking in practice: Cilium and Calico, Gateway API and Envoy, load balancers, ARP, DHCP and the routes that disappear.

Vault in Kubernetes: auto-unseal, Raft, ACL policies, the Vault Secrets Operator and why a sealed Vault breaks everything downstream.

PostgreSQL, CloudNativePG, Amazon RDS and MySQL InnoDB Cluster: replication, cutovers, connection pools and recovery.

Argo CD, Helm and GitLab CI: a safe path from commit to production and the cases where Synced does not mean deployed.

The layer under the cluster: kernels and patching, memory pressure and OOM, virtualization and node maintenance.

Persistent storage for Kubernetes: Longhorn, LINSTOR and DRBD, volume expansion, multipath and migrations without data loss.

Certificates, TLS, secrets and access policies: the security work that keeps production quietly safe.

Running Kafka in production: consumer lag, rebalances, retention and disk capacity.

AWS in production: Amazon EKS and RDS, Multi-AZ cutovers and the limits you meet before the CPU graph does.