Welcome to the Linux Foundation Forum!

Containers vs. Pods and nodeSelector v Taint


I notice that lab 12.1 runs one pod of 4 containers and demonstrates the use of nodeSelector while lab 12.2 runs 8 pods and is demonstrating "taints".

I wish to point out and ask for clarification when lab 12.2 the steps say we will deploy 8 "containers" should it not actually say "pods"?

Am I incorrect to say that the smallest unit that can be deployed to a node is a pod or is it possible that a pod can have containers running on more than one node?

In lab 12.1 we run one pod of 4 containers... so when the nodeSelector is used.. the entire pod of 4 containers is run on either CP or worker depending on which status is used and when there is no nodeSelector then the pod of 4 containers is run on whichever node has fewer containers running.

Would it not be a better demonstration to run 8 pods of 1 container each for Lab 12.1 also so that one can see all 8 pods allocated to CP when nodeSelector is set to vip instead of running just one pod?
Also, then with no nodeSelector used it will better show how the pods are simply distributed accross the two nodes.

Please clarify how containers and Pods are allocated accross nodes..



  • dicalleson

    To help explain where my confusion comes from....
    In Lab 12.1 step 10 it says "Determine where the new containers have been deployed. They should be more evenly spread this time."
    So... if one pod then how is it supposed to "evenly spread" ?


Upcoming Training