2 ms·
Yeah Kubernetes only uses cpusets to pin workload to cores and only when you have this enabled explicitly. Sched_setaffinity is separate mechanism not used by k
by dilyevsky 6y ago
Yeah Kubernetes only uses cpusets to pin workload to cores and only when you have this enabled explicitly. Sched_setaffinity is separate mechanism not used by k8s
99% of typical kubernetes workloads dont need those kind of latency requirements and it maybe be detrimental for their throughput to only use subset of cores (classic throughput vs latency tradeoff).