hard20 pts · first answer countskuberneteslatency

Tail latency at low CPU usage

A CPU model service shows average CPU at 40% of its limit, yet p99 latency has regular spikes. Metrics show high container_cpu_cfs_throttled_seconds. What is happening?

A.Short bursts use the whole CPU quota early in each scheduling period, so the container is throttled until the next oneB.The model has a memory leakC.The network is dropping packetsD.The readiness probe is too frequent

← Same tag, different image More threads, no extra throughput →