Regression in 0.55.0: --spot is only valid with managed nodegroups

[nbusseneau]

Hi,

We use eksctl on cilium/cilium-cli for deploying an EKS cluster and running tests on it via GitHub Actions (link).

First, we download the latest release:

      - name: Install eksctl CLI
        run: |
          curl -LO "https://github.com/weaveworks/eksctl/releases/latest/download/eksctl_$(uname -s)_amd64.tar.gz"
          sudo tar xzvfC eksctl_$(uname -s)_amd64.tar.gz /usr/bin
          rm eksctl_$(uname -s)_amd64.tar.gz

Then, we deploy a cluster without a nodegroup:

      - name: Create EKS cluster without nodegroup
        run: |
          eksctl create cluster \
            --name ${{ env.clusterName }} \
            --tags "usage=${{ github.repository_owner }}-${{ github.event.repository.name }},owner=${{ steps.vars.outputs.owner }}" \
            --without-nodegroup

And finally, we setup a spot managed nodegroup:

      - name: Add managed spot nodegroup
        run: |
          eksctl create nodegroup \
            --cluster ${{ env.clusterName }} \
            --nodes 2 \
            --instance-types "t3.medium,t3a.medium" \
            --node-volume-type gp3 \
            --node-volume-size 10 \
            --managed \
            --spot \
            --node-private-networking

Looking at the list of scheduled runs (running every 6 hours) we can see that the last working run was on Jul 2 around 9 CEST, with the next run around 15 CEST failing at the managed spot nodegroup creation step, along with all subsequent runs:

Error: couldn't create node group filter from command line options: --spot is only valid with managed nodegroups (--managed)

Nothing changed in how we setup EKS clusters between the last working and first failing runs mentioned above.

0.55.0 was released on Jul 2 at 14:27 CEST. I strongly suspect a regression in 0.55.0 introducing the issue. Looking at the list of changes, I feel like #3895 might be related?

I don't have access to my work machine at present but will try the following reproduction steps on Monday:

• Download eksctl 0.54.0 and try to setup a spot managed nodegroup: I expect it will work.
• Download eksctl 0.55.0 and try to setup a spot managed nodegroup: I expect it will not work.

[dz902]

I can reproduce this with 0.55 and 0.56-rc1. This is a fairly common command. It should not slipped under the radar for two releases.

[Callisto13]

Thanks for reporting @nbusseneau, after a quick bisect I found the change came with #3887.

Fix will be along shortly 👍

[nbusseneau]

Can confirm I don't reproduce the issue with 0.54.0, but do with 0.55.0. Editing title.

EDIT: ah sorry, GitHub did not refresh the comment list so I had missed the previous comment 🙈 Thanks!

[cPu1]

Closed by #3930. The fix is out in https://github.com/weaveworks/eksctl/releases/tag/0.56.0-rc.1