CKA logo
Focused certification exam prep
Start practice

CKA Domain 2: Workloads & Scheduling (15%) - Complete Study Guide 2026

TL;DR
  • Domain 2 carries 15% of your CKA score - roughly 2-3 tasks in a 15-20 task exam worth partial credit.
  • You must be able to imperatively create, update, and debug Deployments entirely from the Linux command line in under 2 hours.
  • Scheduling subtopics - taints/tolerations, node affinity, and manual scheduling - are consistently high-value areas in this domain.
  • The CKA is open-book to approved resources inside the exam VM only; external search results are not permitted.

What Domain 2 Actually Tests

Workloads & Scheduling makes up 15% of the CKA Certification exam. That may sound modest compared to Troubleshooting's 30% or Cluster Architecture's 25%, but it is deceptive: workload management is the daily reality of Kubernetes administration, and the scheduling subtopics in this domain surface repeatedly inside other domains as well. If you can't confidently manage Deployments and reason about how the Kubernetes scheduler places Pods, you'll lose points in ways that aren't labeled "Domain 2."

The CNCF and The Linux Foundation designed the CKA to be entirely performance-based. There are no multiple-choice questions. Every task requires you to produce a real outcome - a running Deployment, a correctly constrained Pod, a properly mounted ConfigMap - inside a live cluster environment. For Domain 2, that means you need fluency with kubectl, a mental model of how controllers reconcile desired state, and the ability to write or edit YAML quickly under time pressure.

Domain Weight in Context: At 15%, Domain 2 represents roughly 2-3 tasks in an exam of 15-20 performance-based items. Each task carries partial credit, so a partially correct answer still earns points. Knowing the domain deeply enough to finish most of each task - even if imperfectly - meaningfully improves your final score toward the 66% passing threshold.

For a full picture of how this domain fits alongside the other four, see the CKA Exam Domains 2026: Complete Guide to All 5 Content Areas. Domain 2 sits between the cluster infrastructure focus of Domain 1: Cluster Architecture, Installation & Configuration and the networking concerns of Domain 3: Services & Networking.

Core Topics You Must Master

The official CKA curriculum groups several distinct competencies under Workloads & Scheduling. Below is how those competencies break down into concrete study targets:

Domain 2: Workloads & Scheduling (15%)

Candidates must understand how Kubernetes manages application lifecycle, resource allocation, and Pod placement decisions.

  • Deployments: create, scale, update strategy, rollout history, rollback
  • ReplicaSets: relationship to Deployments, manual selector management
  • StatefulSets and DaemonSets: when to use each, headless services, ordered rollout
  • Jobs and CronJobs: completions, parallelism, schedule syntax, failure handling
  • ConfigMaps and Secrets: creation, injection via env vars and volume mounts
  • Resource requests and limits: CPU/memory units, LimitRange, ResourceQuota
  • Manual scheduling: nodeName, nodeSeletor
  • Node affinity and anti-affinity: required vs. preferred rules
  • Taints and tolerations: NoSchedule, PreferNoSchedule, NoExecute
  • Pod priority and preemption: PriorityClass objects
  • Multi-container Pod patterns: sidecar, ambassador, adapter
  • Init containers: sequencing, failure behavior

Not every topic above carries equal weight in practice. Based on the nature of performance-based Kubernetes administration tasks, scheduling constraints and Deployment lifecycle management tend to generate the most exam-relevant scenarios. The sections that follow go deeper on each high-value cluster.

Deployments, ReplicaSets, and Rolling Updates

A Deployment is the most common way to run stateless workloads in Kubernetes, and it is almost certain to appear in your exam. You should be able to perform every Deployment operation imperatively before you ever open the documentation.

What You Need to Be Able to Do

  • Create a Deployment imperatively: kubectl create deployment nginx --image=nginx:1.25 --replicas=3
  • Scale it: kubectl scale deployment nginx --replicas=5
  • Update the image: kubectl set image deployment/nginx nginx=nginx:1.26
  • Check rollout status: kubectl rollout status deployment/nginx
  • View rollout history: kubectl rollout history deployment/nginx
  • Roll back: kubectl rollout undo deployment/nginx
  • Pause and resume a rollout: used when you need to make multiple changes atomically

The relationship between a Deployment and its ReplicaSets matters too. When you update a Deployment, Kubernetes creates a new ReplicaSet and scales the old one down. Understanding this lets you answer tasks that ask you to inspect why a specific revision exists or manually manipulate replica counts in edge-case scenarios.

Key Takeaway

Generate YAML with --dry-run=client -o yaml rather than writing it from scratch. For example: kubectl create deployment web --image=nginx --dry-run=client -o yaml > web.yaml. This is faster and less error-prone under exam pressure, and the CKA explicitly permits use of kubectl within the exam environment.

StatefulSets and DaemonSets

StatefulSets are used for applications that require stable network identity and ordered deployment - databases, message queues, and similar workloads. You need to understand how they differ from Deployments: Pods are named with ordinal indices, updates are ordered, and they typically pair with a headless Service. DaemonSets ensure exactly one Pod runs on every eligible node - common for log collectors, monitoring agents, and network plugins. Know how to create both, and know how to update a DaemonSet's update strategy.

Scheduling: Manual, Affinity, and Taints

The scheduling subtopics in Domain 2 are among the most nuanced in the entire CKA curriculum. They require both conceptual understanding and precise YAML syntax - a combination that trips up candidates who only studied at a surface level.

Manual Scheduling with nodeName

When the scheduler is bypassed or unavailable, you can force a Pod onto a specific node by setting spec.nodeName directly in the Pod manifest. This is also how you'll answer tasks that ask you to schedule a Pod on a node without using the scheduler at all. Note that nodeName bypasses all affinity rules, taints, and resource checks - it is a direct assignment.

Node Affinity

Node affinity gives you expressive, label-based rules for where Pods should or must be placed:

  • requiredDuringSchedulingIgnoredDuringExecution: Hard rule - Pod won't schedule if no node matches. Used when placement is non-negotiable.
  • preferredDuringSchedulingIgnoredDuringExecution: Soft rule - scheduler tries to honor it but will place the Pod elsewhere if needed.

Exam tasks often combine node affinity with nodeSelector (simpler, key-value label matching) and ask you to distinguish the two. Know the full YAML path: spec.affinity.nodeAffinity.requiredDuringSchedulingIgnoredDuringExecution.nodeSelectorTerms. It is long, and you should look it up in the Kubernetes documentation inside your exam VM rather than memorizing it verbatim.

Taints and Tolerations

Taints are applied to nodes; tolerations are applied to Pods. Together they control which Pods are allowed (or repelled from) specific nodes. The three taint effects you must know:

Taint Effect Behavior Typical Use Case
NoSchedule New Pods without matching toleration will not be scheduled Reserving nodes for specific workloads
PreferNoSchedule Scheduler tries to avoid placing Pods without toleration Soft isolation; fallback if no other nodes available
NoExecute Existing Pods without toleration are evicted; new ones blocked Node maintenance, rolling node drains

A common exam pattern: a node has a taint and a Pod fails to schedule. You're asked to fix it. The fix is adding the correct toleration to the Pod spec - or removing the taint from the node, depending on what the task requires. Practice both.

Resource Requests, Limits, and LimitRanges

Resource management is where scheduling meets operational safety. The scheduler uses requests to decide where to place Pods; the kubelet uses limits to enforce runtime boundaries. Candidates must understand the difference and be able to set both correctly.

Units to Know Cold: CPU is measured in millicores (e.g., 250m = 0.25 cores). Memory uses binary suffixes: 128Mi (mebibytes) and 1Gi (gibibytes) are standard. Using 128M instead of 128Mi is technically valid but refers to different values - a mistake that costs you on a scored task.

Beyond individual Pod specs, you need to work with LimitRange objects (which set default and maximum resource values per namespace) and ResourceQuota objects (which cap total resource consumption across an entire namespace). An exam task might ask you to create a LimitRange that enforces a default memory limit, then create a Pod in that namespace and verify the default was applied.

ConfigMaps and Secrets in Practice

ConfigMaps and Secrets are the primary mechanisms for decoupling configuration from container images. The CKA tests both their creation and their injection into Pods.

Creation Patterns

  • kubectl create configmap app-config --from-literal=ENV=production
  • kubectl create configmap app-config --from-file=config.properties
  • kubectl create secret generic db-secret --from-literal=password=s3cr3t

Injection Patterns

  • Environment variable from ConfigMap: envFrom.configMapRef or individual env.valueFrom.configMapKeyRef
  • Mounted as a volume: each key becomes a file; the value becomes file content
  • Secret as env var: same pattern as ConfigMap but using secretKeyRef

A frequent exam scenario: create a Secret, then modify a running Pod's definition so that the Secret's value is available as an environment variable named something specific. Since Pods are immutable in most respects, you'll need to delete and recreate - which means knowing how to export the existing Pod spec with kubectl get pod <name> -o yaml, edit it, and reapply.

What Domain 2 Tasks Look Like on Exam Day

The CKA exam runs on Kubernetes (verify the current version in the Linux Foundation FAQ before scheduling - it has listed both v1.34 and v1.35 in different locations). You work in a terminal inside a PSI Bridge secure browser. All tasks are practical. Here are representative Domain 2 task patterns:

  • Create a Deployment named web with image nginx:1.25, 4 replicas, in namespace frontend. Update to nginx:1.26 and record the change. Roll back to revision 1.
  • A Pod is in Pending state. Investigate why it is not scheduling. Add the required toleration to make it schedule on the tainted node.
  • Create a Pod that runs only on nodes labeled disk=ssd. Use node affinity, not nodeSelector.
  • Create a ConfigMap with two keys. Mount it as a volume inside a Pod at /etc/config.
  • A Deployment is failing because its containers exceed memory limits. Edit the Deployment to increase the memory limit to 256Mi.
  • Create a CronJob that runs echo hello every 5 minutes, with a concurrencyPolicy of Forbid.

Notice that nearly every task above requires knowing not just the concept but also the specific YAML path and the right imperative command to get there quickly. The 2-hour time limit rewards candidates who can execute without hesitation. For honest context on overall exam difficulty, see How Hard Is the CKA Exam? Complete Difficulty Guide 2026.

Open-Book Strategy for Domain 2: The exam permits the official Kubernetes documentation inside the exam VM. For scheduling-related YAML - especially node affinity's verbose selector paths - navigating directly to the "Assign Pods to Nodes" documentation page is faster than typing from memory. Practice finding these pages in under 30 seconds during your preparation.

You can sharpen your speed on these patterns using the practice environment at CKA Exam Prep practice tests before sitting the real exam.

Focused Study Plan for Domain 2

Domain 2 does not require the most calendar time - that belongs to Troubleshooting (30%) and Cluster Architecture (25%) - but it rewards early, deliberate practice because the scheduling concepts reappear in those heavier domains. A reasonable allocation in a 6-8 week prep schedule:

Week 2

Deployments and Workload Controllers

  • Create, scale, update, and roll back Deployments 10+ times from the CLI
  • Create a StatefulSet with a headless Service; verify ordinal pod names
  • Create a DaemonSet; verify it runs on every node
  • Write and trigger a Job; write a CronJob with Forbid concurrency policy
Week 3

Scheduling and Resource Management

  • Apply and remove taints; write tolerations; verify scheduling behavior
  • Practice required and preferred node affinity rules with real label sets
  • Create LimitRange and ResourceQuota objects; test with non-compliant Pods
  • Work through ConfigMap and Secret injection in both env var and volume-mount patterns

Tie your practice to the official docs. Each time you complete a task, find the corresponding documentation page and note which section you'd reference under time pressure. The two Killer.sh simulator attempts included with your exam registration are ideal for pressure-testing your speed on Domain 2 scenarios before exam day. The exam fee of $445 includes both those simulator attempts and one free retake - details in the CKA Certification Cost 2026: Complete Pricing Breakdown.

For a complete week-by-week schedule covering all five domains, the CKA Study Guide 2026: How to Pass on Your First Attempt provides a structured framework you can adapt to your current experience level. Additional CKA Training resources can supplement your hands-on practice if you're newer to Kubernetes administration.

After you earn your certification, Domain 2 fluency is directly relevant to roles that involve managing containerized workloads at scale. Explore what the credential means for your career in the CKA Salary Guide 2026: Complete Earnings Analysis and browse the types of positions it opens in the CKA Jobs overview. You can also start building your confidence now with the CKA practice exam simulator.

Frequently Asked Questions

How many tasks in the CKA exam come from Domain 2?

The CKA has approximately 15-20 performance-based tasks in total. Domain 2 carries 15% of the exam weight, which corresponds to roughly 2-3 tasks. Because partial credit is awarded per task, you can earn meaningful points even if you don't complete a Domain 2 task perfectly - so never skip a task entirely.

Do I need to memorize YAML for scheduling topics like node affinity?

Not verbatim. The CKA is open-book to approved resources inside the exam VM, including official Kubernetes documentation. Node affinity YAML paths are verbose, and finding them in the docs during the exam is an accepted and practical strategy. What you should be able to do without looking: understand the concept well enough to navigate to the right documentation page in under 30 seconds.

What is the difference between a taint and node affinity?

Taints are applied to nodes to repel Pods - they push workloads away unless a Pod explicitly tolerates the taint. Node affinity is applied to Pods to attract them toward specific nodes - it pulls workloads toward nodes with matching labels. They solve opposite sides of the same placement problem and are often used together in production environments.

Are ConfigMaps and Secrets likely to appear in the exam?

Yes. ConfigMap and Secret tasks appear frequently in the CKA because they reflect real administrative work. Expect tasks that require creating them from literals or files and then injecting them into Pods either as environment variables or volume mounts. Practice both injection patterns until they feel mechanical - the syntax differences between configMapKeyRef and secretKeyRef are small but exam-significant.

How does Domain 2 connect to the other CKA domains?

Workloads & Scheduling concepts appear inside other domains in practice. Troubleshooting tasks (Domain 5, 30%) frequently involve diagnosing Pods that won't schedule due to taints or resource constraints. Services & Networking tasks (Domain 3, 20%) involve Services fronting Deployments. Mastering Domain 2 early creates a foundation that speeds up your work across the entire exam. See the full picture in the CKA Exam Domains 2026: Complete Guide to All 5 Content Areas.

Ready to pass your CKA exam?

Put this into practice with free CKA questions across every exam domain.