how to use kubectl top node in kubernetes python?

Question

UPDATE

Regarding space, I realized I can mount the root path '/' into the container and use this piece of code to get my stats:

import shutil

total, used, free = shutil.disk_usage("/")

print("Total: %d GiB" % (total // (2**30)))
print("Used: %d GiB" % (used // (2**30)))
print("Free: %d GiB" % (free // (2**30)))

Still Looking for a way to do it through Kubernetes itself, though.

Original Question

I'm building a service that part of its function is to monitor system resources for my kubernetes cluster (not specific pods - the entire machine the node runs on) And I realized that kubectl top node is a good way to get that information (excluding storage).

Is there a way to get this information using the kubernetes python package?

I've tried to find a solution in the package documentation and realized that action is not there. (maybe under a different name? I couldn't find it)
Also found this issue on Github which partially solves my problem but I'm still looking for something simpler.

My question: How can I check for system resources like Memory, storage, and CPU usage for my kubernetes node through the Kubernetes Python package?

Answer 1

Resource usage metrics, such as pod or node CPU and memory usage, are available in Kubernetes through the Metrics API.
You can access the Metrics API using kubectl get --raw command, e.g.:

kubectl get --raw /apis/metrics.k8s.io/v1beta1/nodes

More examples can be found in the Metrics API documentation.

---

Using Kubernetes Python Client you should be able to check memory and cpu usage of Kubernetes nodes, just like with the kubectl top nodes command:
NOTE: Memory is in kilobytes but can easily be converted to megabytes.

#!/usr/bin/python
# Script name: usage_info.py

from kubernetes import client, config

config.load_kube_config()

api = client.CustomObjectsApi()
k8s_nodes = api.list_cluster_custom_object("metrics.k8s.io", "v1beta1", "nodes")


for stats in k8s_nodes['items']:
    print("Node Name: %s\tCPU: %s\tMemory: %s" % (stats['metadata']['name'], stats['usage']['cpu'], stats['usage']['memory']))

And sample output:

$ ./usage_info.py
Node Name: node01     CPU: 82845225n  Memory: 707408Ki
Node Name: node02     CPU: 99717207n  Memory: 613892Ki
Node Name: node03     CPU: 74841362n  Memory: 625316Ki

In terms of storage usage, I think it should be checked in a different way as this information isn't available at the /apis/metrics.k8s.io/v1beta1/nodes endpoint.