SRE Skills to Know & Practical Training Path

Whether a company runs on AWS, Microsoft Azure, or Google Cloud, the underlying SRE building blocks are identical—only the service names change:

In interviews, you will almost always be asked: "A Kubernetes Pod is not starting. Which commands do you run?" Memorize this exact sequence:

# Step 1: List all pods and check their STATUS (Running, Pending, CrashLoopBackOff, OOMKilled)
kubectl get pods -A

# Step 2: Inspect detailed events (shows why a pod failed to schedule, mount a secret, or pull an image)
kubectl describe pod <pod-name>

# Step 3: Read the application logs inside the container
kubectl logs <pod-name>

# Step 4: If the container just crashed and restarted, check the PREVIOUS crash logs!
kubectl logs <pod-name> --previous

# Step 5: Check how much CPU and Memory each pod is currently using
kubectl top pods --sort-by=memory

# Step 6: Undo a bad deployment immediately (Rollback to previous working version)
kubectl rollout undo deployment/<deployment-name>