Learn · Workloads

Workloads

Operate controllers, not pods. Learn how Deployments roll, why ReplicaSets exist, and how probes become gates or weapons depending on semantics.

What this module covers

Controller mechanics and operational gates.

Deployments & ReplicaSets: rollout semantics, revision history, and how progress is measured.
Probes: readiness as a traffic gate; liveness as a kill switch; startupProbe as warm-up protection.
Failure states: CrashLoopBackOff, OOMKilled, and rollouts that stall under availability constraints.

Mechanisms first; ceremony as mnemonic.

Text

The Dark Arts of Rollout Safety

Reversibility, staged risk, and how to keep change from becoming an incident.

Text

Probes: Liveness, Readiness, and the Test of Worthiness

Probe semantics that prevent silent outages and restart storms.

Atlas

Deployment Rollout Stalled

Read gates, prove the blocker, then choose rollback vs fix-forward safely.

Repeat the diagnostic sequence until it is reflex.

Lab

CrashLoopBackOff Triage

Events, logs, exit codes, probes, and smallest-safe fixes.

Lab

Rollout Stalled (ProgressDeadlineExceeded)

Deployment conditions, ReplicaSets, readiness gates, and safe rollback posture.

Lab

Probe Semantics Under Load

Design probes that degrade safely under real latency and contention.

Continue with networking.

Learn

Networking

Services, endpoints, ingress, DNS, and the edge-to-pod proof chain.

Topic map

Networking (Archive)

Curated readings and practice links across routing and boundaries.