kops update -f

[krisnova]

Following up on #267:

It looks like we have a use case for

kops update -f

This would feel similar to kops create -f with the only difference being that it is intended to manage deltas and not create a cluster from scratch.

I think this would help users achieve the awesome goal of fully scripted kops management, and following the kubectl paradigm is a shoe-in.

Thoughts anyone?

CC @justinsb @geojaz @yissachar @chrislovecnm @k8s-mirror-aws-misc

[chrislovecnm]

How does this compare to kops replace -f?

[vinayagg]

I came here looking for 'update', found 'replace' instead.
I tested replace, it works perfectly as I think it should.

The only change can be to rename 'kops replace -f' with 'kops update -f' because

• update is more intuitive
• also, I was afraid that replacing with "node" ig yaml will destroy "cluster" and master ig yaml. However, it only replaced node yaml. So it is an update.

[chrislovecnm]

I think we can just alias update to replace with cobra.

[kenden]

I seems update would be more logical, but according to the documentation, update does:
'creates or updates the cloud resources to match the cluster spec.'
(It does not 'update a cloud specification in the registry.`)

I thought the same as @vinayagg, that replace -f would destroy my existing cluster and create a new one. But maybe the problem is simply that there is no documentation about what replace does exactly:
https://github.com/kubernetes/kops/blob/master/docs/commands.md

[pavloos]

It would be great to be able to just have files with deltas and do kops update -f
Let's say I have and additional_policies.yaml containing just changes or additions

spec:
  additionalPolicies:
    master: |
      [
        [...]
      ]
     node: |
      [
        [...]
      ]

and want that merged with existing configuration.
Currently it does not seem possible as kops replace -f complains about missing objects (kind, then apiVersion and so on). Is it supposed to work and I am missing something or replace/update still requires full cluster config file ?

[fejta-bot]

Issues go stale after 90d of inactivity.
Mark the issue as fresh with /remove-lifecycle stale.
Stale issues rot after an additional 30d of inactivity and eventually close.

Prevent issues from auto-closing with an /lifecycle frozen comment.

If this issue is safe to close now please do so with /close.

Send feedback to sig-testing, kubernetes/test-infra and/or @fejta.
/lifecycle stale

[geojaz]

/close

we dunnit!