NodeSelector for Job - use it only if NodeSelector exists

Question

I have a Job Object which shall use a Node Selector to only use Nodes, which have a GPU under the hood. I know to to set it (it gets converted from a string in a python program).

    job = f"""
    apiVersion: batch/v1
    kind: Job
    ....
          nodeSelector:
            sma-gpu-size: {gpu_size}
    """

Our ops team sets this selectors in the next few weeks, but currently when setting the node selector, the service is not able to start.

2022-09-20T07:20:24Z [Warning] 0/35 nodes are available: 2 node(s) had taint {node-role.kubernetes.io/infra: }, that the pod didn't tolerate, 3 node(s) had taint {node-role.kubernetes.io/master: }, that the pod didn't tolerate, 30 node(s) didn't match Pod's node affinity/selector.

Is it somehow possible to use these node_selectors only if they are available, something like this (pseudo yaml)?

    job = f"""
    apiVersion: batch/v1
    kind: Job
    ....
          nodeSelector:
            if_available:
                sma-gpu-size: {gpu_size}
            else:
                Any
    """

Answer 1

It's not, but you can replace the nodeSelector with a nodeAffinity to achieve that.

spec:
  [...]
  affinity:
    nodeAffinity:
      preferredDuringSchedulingIgnoredDuringExecution:
        nodeSelectorTerms:
        - matchExpressions:
          - key: sma-gpu-size
            operator: In
            values:
            - {gpu_size}

From docs:

preferredDuringSchedulingIgnoredDuringExecution: The scheduler tries to find a node that meets the rule. If a matching node is not available, the scheduler still schedules the Pod.

After the label has been added, you can switch to requiredDuringSchedulingIgnoredDuringExecution:

requiredDuringSchedulingIgnoredDuringExecution: The scheduler can't schedule the Pod unless the rule is met. This functions like nodeSelector, but with a more expressive syntax.

or back to nodeSelector.