问题

我有这个查询：

100 * (1 - ((avg_over_time(node_memory_MemFree_bytes[10m]) + avg_over_time(node_memory_Cached_bytes[10m]) + avg_over_time(node_memory_Buffers_bytes[10m])) / avg_over_time(node_memory_MemTotal_bytes[10m])))

然而，它只返回安装了Prometheus的命名空间：

{instance="10.240.0.11:9100", job="kubernetes-service-endpoints", kubernetes_name="node-exporter", kubernetes_namespace="monitoring"}
5.58905365516873
{instance="10.240.0.11:9100", job="node-exporter"}
5.588556605118522
{instance="10.240.0.42:9100", job="kubernetes-service-endpoints", kubernetes_name="node-exporter", kubernetes_namespace="monitoring"}
5.093870850709847
{instance="10.240.0.42:9100", job="node-exporter"}
5.09401539556571
{instance="10.240.0.90:9100", job="kubernetes-service-endpoints", kubernetes_name="node-exporter", kubernetes_namespace="monitoring"}
5.103046564234582
{instance="10.240.0.90:9100", job="node-exporter"}

是否可能有一个类似的查询，查询整个集群，所有节点和命名空间？如果是，怎么做？

100 * (1 - ((avg_over_time(node_memory_MemFree_bytes[10m]) + avg_over_time(node_memory_Cached_bytes[10m]) + avg_over_time(node_memory_Buffers_bytes[10m])) / avg_over_time(node_memory_MemTotal_bytes[10m])))

{instance="10.240.0.11:9100", job="kubernetes-service-endpoints", kubernetes_name="node-exporter", kubernetes_namespace="monitoring"}
5.58905365516873
{instance="10.240.0.11:9100", job="node-exporter"}
5.588556605118522
{instance="10.240.0.42:9100", job="kubernetes-service-endpoints", kubernetes_name="node-exporter", kubernetes_namespace="monitoring"}
5.093870850709847
{instance="10.240.0.42:9100", job="node-exporter"}
5.09401539556571
{instance="10.240.0.90:9100", job="kubernetes-service-endpoints", kubernetes_name="node-exporter", kubernetes_namespace="monitoring"}
5.103046564234582
{instance="10.240.0.90:9100", job="node-exporter"}

答案1

得分: 3

使用作为守护程序安装的 node-exporter，您可以获得整个集群的指标。

要获取整个集群内存使用率（以百分比表示）：

 sum(node_memory_MemTotal_bytes{service="node-exporter"}) -
 sum(node_memory_MemAvailable_bytes{service="node-exporter"})
) / sum(node_memory_MemTotal_bytes{service="node-exporter"})```
结果，例如：
```{} 37.234674067149946```

按节点查看内存使用情况：
```100 * (
 sum by (instance) (node_memory_MemTotal_bytes{service="node-exporter"}) - 
 sum by (instance) (node_memory_MemAvailable_bytes{service="node-exporter"})
) / sum by (instance) (node_memory_MemTotal_bytes{service="node-exporter"})```
结果，例如：
```{instance="x.x.x.x:9100"} 42.51742364002058
{instance="y.y.y.y:9100"} 38.26956501095188
{instance="z.z.z.z:9100"}	36.57150031634585```

特定命名空间的内存使用情况：
```100 * sum(container_memory_working_set_bytes{namespace="my-namespace"}) / 
sum(node_memory_MemTotal_bytes)```
结果，例如：
```{} 4.212481093513011```

<details>
<summary>英文:</summary>

With the node-exporter installed as daemonset, you have the metrics of the entire cluster.

To have the overall cluster memory usage, in percentage:

100 * (
sum(node_memory_MemTotal_bytes{service="node-exporter"}) -
sum(node_memory_MemAvailable_bytes{service="node-exporter"})
) / sum(node_memory_MemTotal_bytes{service="node-exporter"})

Result, for example:

{} 37.234674067149946

Memory usage by node:

100 * (
sum by (instance) (node_memory_MemTotal_bytes{service="node-exporter"}) -
sum by (instance) (node_memory_MemAvailable_bytes{service="node-exporter"})
) / sum by (instance) (node_memory_MemTotal_bytes{service="node-exporter"})

Result, for example:

{instance="x.x.x.x:9100"} 42.51742364002058
{instance="y.y.y.y:9100"} 38.26956501095188
{instance="z.z.z.z:9100"} 36.57150031634585

Memory usage for a specific namespace:

100 * sum(container_memory_working_set_bytes{namespace="my-namespace"}) /
sum(node_memory_MemTotal_bytes)

Result, for example:

{} 4.212481093513011

</details>