Prometheus Statements

Prometheus Statement Syntax

For details about the Prometheus statement syntax, go to the Prometheus official website.

Examples of Using Prometheus Statements

• Example 1: Memory usage of a specified pod in a node (excluding the control node)
  • Define variables:
    • Used memory of the containers in a pod (a pod may contain multiple containers or instances): aom_container_memory_used_megabytes
    • Total memory of the node: aom_node_memory_total_megabytes
  • Query logic:
    • For aom_container_memory_used_megabytes, use the aggregation function sum to calculate the actual used memory of a specified pod under a specified node based on the node IP address and pod ID.
    • For aom_node_memory_total_megabytes, use the aggregation function sum to calculate the total memory of a specified node based on the node IP address.
    • Both of them are filtered by node IP address. Therefore, the obtained metric values have the same metric dimension. (Only the values are different.)
    • The actual memory usage of the pod can be obtained by performing the "/" operation on the values of the preceding two metrics.
  • To query the actual memory usage of the pod, use the following statement:

    sum(aom_container_memory_used_megabytes{podID="****1461-41d8-****-bfeb-fc1213****",nodeIP="***.***.***.***"}) by (nodeIP) / sum(aom_node_memory_total_megabytes{nodeIP="***.***.***.***"}) by (nodeIP)

• Example 2: CPU usage of a specified pod in a node (excluding the control node)
  • Define variables:
    • Used CPU cores of the containers in a pod: aom_container_cpu_used_core
    • Actual total number of CPU cores of the node: aom_node_cpu_limit_core
  • Query logic:
    • For aom_container_cpu_used_core, use the aggregation function sum to calculate the used CPU cores of a specified pod under a specified node based on the node IP address and pod ID.
    • For aom_node_cpu_limit_core, use the aggregation function sum to calculate the total CPU cores of a specified node based on the node IP address.
    • Both of them are filtered by node IP address. Therefore, the obtained metric values have the same metric dimension. (Only the values are different.)
    • The actual memory usage of the pod can be obtained by performing the "/" operation on the values of the preceding two metrics.
  • To obtain the actual CPU usage of the pod, use the following statement:

    sum(aom_container_cpu_used_core{nodeIP="***.***.***.***",podID="****1461-41d8-****-bfeb-***13******"}) by (nodeIP) / sum(aom_node_cpu_limit_core{nodeIP="***.***.***.***"}) by (nodeIP)

• Example 3: Requested memory of a pod/Allocable memory of the node where the pod is located
  • Define variables:
    • Memory allocated to the containers in a pod: aom_container_memory_request_megabytes
    • Total memory of the node: aom_node_memory_total_megabytes
  • Query logic:
    • For aom_container_memory_request_megabytes, use the aggregation function sum to calculate the allocated memory of a specified pod under a specified node based on the node IP address and pod ID.
    • For aom_node_memory_total_megabytes, use the aggregation function sum to calculate the total memory of a specified node based on the node IP address.
    • Both of them are filtered by node IP address. Therefore, the obtained metric values have the same metric dimension. (Only the values are different.)
    • The actual memory usage of the pod can be obtained by performing the "/" operation on the values of the preceding two metrics.
  • To obtain the actual memory allocation ratio of the pod, use the following statement:

    sum(aom_container_memory_request_megabytes{podID="****1461-41d8-4403-****-f***35*****",nodeIP="***.***.***.***"}) by (nodeIP) / sum(aom_node_memory_total_megabytes{nodeIP="***.***.***.***"}) by (nodeIP)

• Example 4: Requested CPU cores of a pod/Allocable CPU cores of the node where the pod is located
  • Define variables:
    • CPU cores allocated to the containers in the pod: aom_container_cpu_limit_core
    • CPU cores allocated to the node: aom_node_cpu_limit_core
  • Query logic:
    • For aom_container_cpu_limit_core, use the aggregation function sum to calculate the CPU cores allocated to a specified pod under a specified node based on the node IP address and pod ID.
    • For aom_node_cpu_limit_core, use the aggregation function sum to calculate the total CPU cores of a specified node based on the node IP address.
    • Both of them are filtered by node IP address. Therefore, the obtained metric values have the same metric dimension. (Only the values are different.)
    • The actual CPU usage of the pod can be obtained by performing the "/" operation on the values of the preceding two metrics.
  • To obtain the actual CPU allocation ratio of the pod, use the following statement:

    sum(aom_container_cpu_limit_core{podID="*****461-41d8-****-bfeb-****135*****",nodeIP="***.***.***.***"}) by (nodeIP) / sum(aom_node_cpu_limit_core{nodeIP="***.***.***.***"}) by (nodeIP)

Common Prometheus Commands

Table 1 lists the common Prometheus commands for querying metrics. You can modify parameters such as the IP address and ID based on site requirements.