add NODE info for kubectl top pod output

[qyecst]

What would you like to be added:

Hi team, I want to add NODE info for kubectl top pod output like

NAME                   NODE     CPU(cores)   MEMORY(bytes)   
db-5977df8496-p8s2f    node1    1m           57Mi
one-d48cbd8b8-znqcx    node2    2m           1Mi
two-55445bbdb9-bt72c   node3    5m           2Mi

Why is this needed:
Convenient to view the pod status of the specified node: kubectl top pod |grep node-name

Convenient to check the load and node distribution of pods: kubectl top pod -l app=label

When sudden surge in node load, convenient to check the status of the corresponding pods.

---

The current output does not include node information. When we need to handle high-load pods or pods on high-load nodes, we need to use the kubectl get command to obtain the node where the pod is located.

This involves many additional commands and operations, such as handling and merging information from get pod and get podmetrics.

Using Prometheus for information aggregation is a solution, but it still takes a lot of time to pass pod names between Prometheus and the terminal when you need to operate on pods, especially when there are many pods involved.

[qyecst]

I have worked on this feature and ready to contribute my implementation via a pull request.

[qyecst]

/sig cli

[Ritikaa96]

@kubernetes/sig-cli-kubectl-maintainers
tagging for triaging

[googs1025]

Maybe we should listen to some use cases. I don't know the necessary relationship of adding node info. Although feature is not big, we should not provide feature without describing clearly.

[qyecst]

What would you like to be added:

Hi team, I want to add NODE info for kubectl top pod output like

NAME                   NODE     CPU(cores)   MEMORY(bytes)   
db-5977df8496-p8s2f    node1    1m           57Mi
one-d48cbd8b8-znqcx    node2    2m           1Mi
two-55445bbdb9-bt72c   node3    5m           2Mi

Why is this needed: Convenient to view the pod status of the specified node: kubectl top pod |grep node-name

Convenient to check the load and node distribution of pods: kubectl top pod -l app=label

When sudden surge in node load, convenient to check the status of the corresponding pods.

The current output does not include node information. When we need to handle high-load pods or pods on high-load nodes, we need to use the kubectl get command to obtain the node where the pod is located.

This involves many additional commands and operations, such as handling and merging information from get pod and get podmetrics.

Using Prometheus for information aggregation is a solution, but it still takes a lot of time to pass pod names between Prometheus and the terminal when you need to operate on pods, especially when there are many pods involved.

@kubernetes/sig-cli-kubectl-maintainers
Hi, i added some use cases, please tagging for triaging

[mpuckett159]

/triage accepted
With the caveat that this new view is triggered by a flag such as -o wide which is similar to the get subocmmand flag.

[Ritikaa96]

The implementation will be similar to kubernetes/kubernetes#126529
I'll like to contribute this feature.

[googs1025]

I have worked on this feature and ready to contribute my implementation via a pull request.

Maybe @qyecst has already invested himself in this work. Are you still involved in it?

[qyecst]

@googs1025 Yes, but needs to be modified to use a flag like -o wide, probably this weekend.

[googs1025]

/assign @qyecst