如何在kubectl resource-capacity输出中追加Namespace列
2026-8-20
问题
我通过以下kubectl插件命令生成资源利用率输出：
./kubectl get namespaces | awk '{ print $1 }' | while read x; do ./kubectl resource-capacity --sort cpu.util --namespace $x --util --pod-count --pods; done
执行过程中出现权限错误：
Error getting Pod Metrics: pods.metrics.k8s.io is forbidden: User "u-567" cannot list resource "pods" in API group "metrics.k8s.io" in the namespace "NAME" For this to work, metrics-server needs to be running in your cluster
同时得到了如下资源利用率输出：
NODE POD CPU REQUESTS CPU LIMITS CPU UTIL MEMORY REQUESTS MEMORY LIMITS MEMORY UTIL POD COUNT * * 16550m (16%) 45060m (45%) 0Mi (0%) 38174Mi (9%) 33492Mi (8%) 0Mi (0%) 16/1540 ip-10-11-7-77.eu-west-1.compute.internal * 6150m (76%) 14160m (177%) 139m (1%) 16784Mi (53%) 9632Mi (30%) 3126Mi (9%) 5/110 ip-10-11-7-77.eu-west-1.compute.internal anomaly-detector-water-transformer-default-msxhv-deploymenrnv85 1035m (12%) 3040m (38%) 46m (0%) 2138Mi (6%) 2376Mi (7%) 300Mi (0%) ip-10-11-7-77.eu-west-1.compute.internal anomaly-detector-blades-transformer-default-gmvb2-deploymezzt5r 1035m (12%) 3040m (38%) 45m (0%) 2138Mi (6%) 2376Mi (7%) 292Mi (0%) ip-10-11-7-77.eu-west-1.compute.internal leadec-uk-inference-m1-predictor-default-zvxkb-deployment-8mscd 1035m (12%) 3040m (38%) 43m (0%) 2138Mi (6%) 2376Mi (7%) 277Mi (0%) ip-10-11-7-77.eu-west-1.compute.internal datascience-0 2010m (25%) 2000m (25%) 8m (0%) 8232Mi (26%) 128Mi (0%) 2259Mi (7%) ip-10-11-7-77.eu-west-1.compute.internal anomaly-detector-leadec-predictor-default-tqjsg-deployment4bqdc 1035m (12%) 3040m (38%) 0Mi (0%) 2138Mi (6%) 2376Mi (7%) 0Mi (0%) ip-10-11-7-134.eu-west-1.compute.internal * 2070m (25%) 6080m (76%) 88m (1%) 4276Mi (13%) 4752Mi (15%) 597Mi (1%) 2/110 ip-10-11-7-134.eu-west-1.compute.internal leadec-uk-inference-m1-transformer-default-84tk6-deploymeng4qjf 1035m (12%) 3040m (38%) 45m (0%) 2138Mi (6%) 2376Mi (7%) 298Mi (0%) ip-10-11-7-134.eu-west-1.compute.internal leadec-uk-synthetic-f65535db1d5e1078fca102b9506391d8-deplop74ww 1035m (12%) 3040m (38%) 44m (0%) 2138Mi (6%) 2376Mi (7%) 299Mi (0%) ip-10-11-7-123.eu-west-1.compute.internal * 1035m (12%) 3040m (38%) 55m (0%) 2138Mi (6%) 2376Mi (7%) 317Mi (1%) 1/110 ip-10-11-7-123.eu-west-1.compute.internal leadec-uk-synthetic-58a5440150937e2198bfdd610a109b95-deplomfn97 1035m (12%) 3040m (38%) 55m (0%) 2138Mi (6%) 2376Mi (7%) 317Mi (1%) ip-10-11-7-190.eu-west-1.compute.internal * 1085m (13%) 3540m (44%) 50m (0%) 2148Mi (6%) 2476Mi (7%) 327Mi (1%) 2/110 ip-10-11-7-190.eu-west-1.compute.internal leadec-uk-synthetic-year3-m4-predictor-default-ttdkp-deplovvtj6 1035m (12%) 3040m (38%) 47m (0%) 2138Mi (6%) 2376Mi (7%) 267Mi (0%) ip-10-11-7-190.eu-west-1.compute.internal random-experiment-random-568ff69c58-j7lx4 50m (0%) 500m (6%) 3m (0%) 10Mi (0%) 100Mi (0%) 60Mi (0%) ip-10-11-7-44.eu-west-1.compute.internal * 1035m (12%) 3040m (38%) 46m (0%) 2138Mi (6%) 2376Mi (7%) 268Mi (0%) 1/110 ip-10-11-7-44.eu-west-1.compute.internal leadec-uk-synthetic-year4-m4-predictor-default-7d9cb-deplofc5kf 1035m (12%) 3040m (38%) 46m (0%) 2138Mi (6%) 2376Mi (7%) 268Mi (0%) ip-10-11-7-57.eu-west-1.compute.internal * 1035m (12%) 3040m (38%) 46m (0%) 2138Mi (6%) 2376Mi (7%) 290Mi (0%) 1/110 ip-10-11-7-57.eu-west-1.compute.internal anomaly-detector-leadec-transformer-default-7j42s-deploymeqhxtd 1035m (12%) 3040m (38%) 46m (0%) 2138Mi (6%) 2376Mi (7%) 290Mi (0%) ip-10-11-7-113.eu-west-1.compute.internal * 1035m (25%) 3040m (76%) 43m (1%) 2138Mi (13%) 2376Mi (15%) 298Mi (1%) 1/110 ip-10-11-7-113.eu-west-1.compute.internal leadec-uk-synthetic-m4-transformer-default-9d4pz-deploymenhtfsv 1035m (25%) 3040m (76%) 43m (1%) 2138Mi (13%) 2376Mi (15%) 298Mi (1%) ip-10-11-7-168.eu-west-1.compute.internal * 0Mi (0%) 0Mi (0%) 0Mi (0%) 0Mi (0%) 0Mi (0%) 0Mi (0%) 0/110
我希望在上述输出的每一行中追加Namespace作为一列，请问该如何实现？
解决方案
修改原循环命令，在处理每个命名空间的输出时为每行添加当前命名空间列，同时确保表头只输出一次：
./kubectl get namespaces | awk '{ print $1 }' | while read x; do # 捕获命令输出，屏蔽权限错误 output=$("./kubectl" resource-capacity --sort cpu.util --namespace "$x" --util --pod-count --pods 2>/dev/null) if [ -n "$output" ]; then # 首次输出时添加表头的NAMESPACE列 if [ -z "$header_printed" ]; then echo "$output" | head -1 | awk '{print $0, "NAMESPACE"}' header_printed=1 # 输出剩余行并追加命名空间 echo "$output" | tail -n +2 | awk -v ns="$x" '{print $0, ns}' else # 非首次输出跳过表头，直接追加命名空间 echo "$output" | tail -n +2 | awk -v ns="$x" '{print $0, ns}' fi fi done
关键细节说明
2>/dev/null：将权限错误信息重定向到空设备，避免干扰正常输出header_printed变量：确保表头仅打印一次，不会重复出现awk -v ns="$x"：将当前循环的命名空间作为变量传递给awk，追加到每行末尾- 空输出判断：过滤没有权限或无资源的命名空间，避免产生无效内容
另外，针对权限错误，需要给用户u-567添加pods.metrics.k8s.io的list权限，可通过以下RBAC配置实现：
apiVersion: rbac.authorization.k8s.io/v1 kind: ClusterRole metadata: name: metrics-reader rules: - apiGroups: ["metrics.k8s.io"] resources: ["pods"] verbs: ["list", "get", "watch"] --- apiVersion: rbac.authorization.k8s.io/v1 kind: ClusterRoleBinding metadata: name: u-567-metrics-reader subjects: - kind: User name: u-567 apiGroup: rbac.authorization.k8s.io roleRef: kind: ClusterRole name: metrics-reader apiGroup: rbac.authorization.k8s.io
内容的提问来源于stack exchange，提问作者utkarsh saraf
相关产品推荐
相关产品推荐
