Cannot run k6-operator "exec /manager: exec format error"

[danielfigueiredo]

Brief summary

I've tried installing the k6-operator both via helm (latst version 3.5.0.) or via make deploy using latest master. Both way produce the exact same error.

The manager pod fails to start with error, however kube-rbac-proxy start just fine.

I'm running on a mac aarch64-darwin system

k6-operator version or image

latest master

Helm chart version (if applicable)

3.5.0

TestRun / PrivateLoadZone YAML

apiVersion: k6.io/v1alpha1
kind: TestRun
metadata:
  name: k6-sample
spec:
  parallelism: 4
  script:
    configMap:
      name: crocodile-stress-test
      file: single-request.js

Other environment details (if applicable)

No response

Steps to reproduce the problem

1. start brand new minikube cluster
2. install k6-operator (either way make deploy of master or helm install .... as per README.md)
3. Check k6-operator-controller-manager, status is CrashLoopBackOff
4. Look up logs for the manager pod to find the following error:

manager exec /manager: exec format error  
Stream closed EOF for k6-operator-system/k6-operator-controller-manager-645f98f9d9-487fz                                                 

No errors during install or anything, it only fails booting up.

Expected behaviour

The resources required by the k6-operator bootstraps fine and reliably when using later or main versions.

Actual behaviour

manager status on fresh install on fresh cluster is CrashLoopBackOff with error manager exec /manager: exec format error

[danielfigueiredo]

Not sure if related, but https://github.com/grafana/k6-operator/blob/main/config/manager/manager.yaml#L30

[danielfigueiredo]

Tried spinning up a local cluster using kind instead of minikube, but same error:

k6.operator.mov

[yorugac]

Hi @danielfigueiredo!

I'm running on a mac aarch64-darwin system

This is ARM arch, right? If so, we have an issue for that here #187
And there has been a PR recently:
#380

I'll try to ping the author there as it's been awhile. But he has the image built there: could you please check if it works for you? It'd be helpful to confirm for your platform.

[audacioustux]

Hi, @danielfigueiredo you may try this repo (until it's merged to upstream), and pls let me know if it works out for you.
https://github.com/audacioustux/k6-operator

Example TestRun manifest:

apiVersion: k6.io/v1alpha1
kind: TestRun
metadata:
  name: closed-test
spec:
  parallelism: 5
  starter:
    image: ghcr.io/audacioustux/k6-operator:starter-latest
    imagePullPolicy: Always
  runner:
    image: ghcr.io/audacioustux/k6-operator:runner-latest
    imagePullPolicy: Always
  arguments: --out cloud
  script:
    configMap:
      name: closed-lt
      file: index.js

[danielfigueiredo]

@audacioustux your branch + docker img does work for me! 🎉

The only thing I couldn't test was the install via helm, I just ran make deploy

Thanks all

[yorugac]

Perfect; thank you, @danielfigueiredo, for checking!

[danielfigueiredo]

I think it makes sense for me to close this issue given we are already tracking the error here #187 ?

I'll subscribe and follow up on the other PR/Issues 😄