Support fallback paths for custom columns

[sule26]

Hello again!

I'm starting to feel like I'm asking for a lot. 😅

Would it be possible to define fallback paths for columns?

For example, I have some clusters using Karpenter, so I created a column that displays the karpenter.sh/nodepool label. However, I also have a single node group managed by AWS Auto Scaling, which does not have that label and instead uses eks.amazonaws.com/nodegroup.

Because the configured path does not exist for those nodes, the column currently displays <none>. It would be really useful if columns could support fallback paths, where the first available value is used.

Something like:

columns = [
  { name = "NAME",          builtin = "NAME" },
  { name = "STATUS",        builtin = "STATUS", width = 10, align = "center" },
  { name = "TAINTS",        builtin = "TAINTS", align = "center" },
  { name = "NODEPOOL",      path = ["/metadata/labels/karpenter.sh~1nodepool", "/metadata/labels/eks.amazonaws.com~1nodegroup"] },
  { name = "CAPACITY_TYPE", path = "/metadata/labels/karpenter.sh~1capacity-type", align = "center" },
  { name = "ARCH",          path = "/metadata/labels/kubernetes.io~1arch" },
  { name = "ZONE",          path = "/metadata/labels/topology.kubernetes.io~1zone" },
  { name = "INSTANCE",      path = "/metadata/labels/node.kubernetes.io~1instance-type" },
  # ...
]

The paths would be evaluated in order, and the first one that exists would be displayed.

Another possible syntax could be a fallback field:

{ name = "NODEPOOL",
  path = "/metadata/labels/karpenter.sh~1nodepool",
  fallback = "/metadata/labels/eks.amazonaws.com~1nodegroup"
}

This would be particularly useful for Kubernetes resources where the same information can come from different controllers or implementations.

For example, in my case:

• Karpenter nodes → karpenter.sh/nodepool
• EKS managed node group → eks.amazonaws.com/nodegroup

This would also make custom column configurations much more reusable across clusters with slightly different setups.

How it is now:

[nklmilojevic]

Thank you for the suggestion and the clear example! Please keep the feedback coming. It is always valuable, and reasonable features that fit sofka will always find their way in. :)

This is a useful case, and I agree with the scope. I prefer your first option: let path accept either one string or an ordered list of paths. That keeps existing configurations valid and supports more than one fallback.

The first value that exists and is not null will be used. Empty strings, 0, and false will remain valid values. If no path provides a value, the column will show <none>. Formatting and sorting will use the selected value.

I will create an issue to track this and start the implementation. Thanks again for helping improve sofka!

[nklmilojevic]

Tracked in #739

[sule26]

Awesome! Thanks! I'll be looking forward to it