Affinity selects or prefers placement; taints repel Pods unless tolerated. A toleration permits placement but does not force it.
Before you start
You should understand Pods, Deployments and Services. Read desired configuration separately from observed cluster state. Use a development cluster when trying changes, and inspect events and status rather than assuming that an accepted manifest means the workload is ready to serve traffic.
The practical goal is to reason through this situation: A specialized workload tolerates a dedicated node taint and also selects the intended node group. Read the walkthrough first, then try the interview exercise before opening its answer. The important part is explaining the decision and its consequences, rather than remembering a definition alone.
Step-by-step walkthrough
Step 1: Separate permission and selection
A toleration allows a tainted node but does not force placement there.
Step 2: Apply required selection
Use appropriate affinity or selectors for mandatory node characteristics.
Step 3: Check available capacity
Strict placement reduces the set of nodes available for rollout and recovery.
Worked scenario
A specialized workload tolerates a dedicated node taint and also selects the intended node group.
A GPU workload tolerates the dedicated-node taint but can still land on an ordinary node unless another requirement selects GPU nodes. With strict selection added, insufficient dedicated capacity can leave it pending. Preferences and requirements have different tradeoffs; specify which the application actually needs.
Common mistake
A toleration alone can allow scheduling on ordinary nodes too.
Verify the behavior
Test eligible and ineligible nodes plus lack of capacity.
Interview exercise
Guarantee a placement requirement.
Answer and reasoning
Use the appropriate required selection rule, then ensure enough eligible capacity exists for rollout and failure.
Continue learning
Compare the scenario with the Kubernetes interview questions and test your understanding with the Kubernetes MCQs. For terminology and implementation details, consult the reference material.