Configuring Pod-level Labels
Important | The pod-level labelling configuration is applicable only when the SMF is deployed on a bare metal server. |
Use the following sample configuration to configure the SMF pod layout when the virtual machine is short of CPU and memory resources.
config
endpoint protocol
labels key label_key value label_value
cpu { max-process process_thread_count | request resource_request_number}
memory { limit max_resource_limit | request resource_request_number}
end
NOTES:
-
labels key label_key value label_value : Specify the K8 node affinity label key and value.
label_key and label_value accept alphanumeric characters. For example, the key can be smi.cisco.com/protocol.
ImportantThe pod-level configuration takes precedence over the layered node-level configuration, that is, at the protocol, service, or session level configuration. -
cpu { max-process process_thread_count | request resource_request_number } : Enables the K8 pod CPU configuration.
-
max-process process_thread_count : Specify the maximum number of parallel OS threads to use. process_thread_count must be an integer in the range of 1-32.
-
request resource_request_number : Specify the CPU resource request in millicores. resource_request_number must be an integer in the range of 100-1000000.
-
-
memory { limit max_resource_limit | request resource_request_number } : Enables the K8 pod memory configuration.
-
limit max_resource_limit : Specify the maximum number of used memory resources in megabytes. max_limit must be an integer in the range of 100-200000.
-
request resource_request_number : Specify the memory resource request in megabytes. request_number must be an integer in the range of 100-200000.
-
Use the following table for node-level labelling.
Node |
OAM |
Protocol |
CDL |
SMF |
---|---|---|---|---|
Node 1 |
Yes |
Yes |
Yes |
No |
Node 2 |
Yes |
Yes |
Yes |
No |
Node 3 |
Yes |
No |
No |
Yes |
Node 4 |
No |
No |
No |
Yes |