<aside> ☸️

Буллеты про Kubernetes: деплой, workloads, health checks, graceful shutdown, resources, autoscaling, secrets, сеть, observability, operators и platform engineering.

</aside>

Деплой и рабочие нагрузки

• Deployed and operated X microservices on Kubernetes across X clusters serving Xk RPS.
• Migrated X services from VMs to Kubernetes, cutting deployment time from Xmin to Xmin and infra cost X%.
• Defined Deployments, Services and Ingress for X applications with declarative manifests under version control.
• Implemented rolling updates with maxSurge/maxUnavailable tuning, achieving zero-downtime releases across X deploys/week.
• Built StatefulSets for X stateful components with persistent volumes and ordered rollout guarantees.
• Ran X CronJobs on Kubernetes with concurrency policies and failure alerting, replacing X host-level crons.
• Configured pod disruption budgets across X workloads, keeping availability above X% during node maintenance.
• Introduced init containers and sidecars for X cross-cutting concerns, removing duplicated logic from X services.
• Set up X namespaces with resource quotas and limit ranges, preventing X noisy-neighbour incidents per quarter.
• Migrated X workloads to a new cluster version across X minor upgrades with no customer-visible downtime.

Здоровье, готовность и graceful shutdown

• Implemented liveness, readiness and startup probes for X Go services, eliminating traffic to unready pods.
• Built graceful shutdown handling SIGTERM with an Xs drain window, removing X dropped requests per deploy.
• Fixed X false-positive restarts by separating liveness from readiness and tuning probe thresholds.
• Added dependency-aware readiness (database, cache, broker), cutting failed requests after rollout X%.
• Introduced preStop hooks with connection draining, eliminating X% of 502 errors during rollouts.
• Reduced pod startup time from Xs to Xs, cutting rollout duration for a X-replica deployment from Xmin to Xmin.

Ресурсы и автоскейлинг