Grading Checklist

• Identifies node memory pressure as the trigger for kubelet-initiated evictions.
• Explains the kubelet eviction mechanism: soft and hard eviction thresholds.
• Checks kubectl describe node for conditions and allocated resources.
• Identifies pods without resource requests/limits as the key contributing factor.
• Explains pod eviction priority: BestEffort pods evicted first, then Burstable, then Guaranteed.
• Recognizes the rescheduling loop: pods without requests land back on the same node.
• Recommends setting memory requests and limits on all workloads.
• Suggests using LimitRange to enforce default requests/limits per namespace.
• Mentions ResourceQuota as a guardrail against namespace overcommitment.
• Considers enabling cluster autoscaler or adding node capacity.
• Recommends checking for memory leaks in the workloads consuming the most memory.
• Notes the importance of monitoring actual vs. requested memory to detect overcommitment.