Installation stalled waiting for modules to be ready..

[mikebaldry]

It has done kalm-operator and cert-manager, and now it just sits there doing nothing. How can I help diagnose this issue?

[infrar]

hey @mikebaldry

After installing kalm-operator and cert-manager, I believe the script will try to install istio. Depending on network conditions this can sometimes take up to 5-10 minutes.

One way to diagnose the status of installing operators is to look at the status of namespaces with kubectl, for example:

kubectl get ns
kubectl get all -n istio-system

Example output from my dev machine:

From here, you can try to identify whether a pod is stuck, and further diagnose by examining pods.

[mikebaldry]

Yes, I see in the titlebar it's waiting for the istio stuff to appear. After inspecting my cluster, I don't have an istio-system at all. It has been running for hours now, so I guess I'll stop it and run again, see what happens?

When I look at the logs of kalm operator, I see this error repeated..

controllers.KalmOperatorConfig install certManager error. {"kalmoperatorconfig": "kalm-operator/kalmoperatorconfig-sample", "error": "CustomResourceDefinition.apiextensions.k8s.io \"certificaterequests.cert-manager.io\" is invalid: spec.conversion.webhookClientConfig: Required value: required when strategy is set to Webhook, but not allowed because the CustomResourceWebhookConversion feature is disabled"} [kalm-operator-6fd85d694-hl6gw manager] github.com/go-logr/zapr.(*zapLogger).Error

[davidqhr]

It seems your k8s cluster version is low, some feature is not enabled by default. Can you try kalm on a k8s cluster which version is not less than 1.15?

[mikebaldry]

Ah yes I'm on 1.14. it suggests in the docs it's supported. What do I need to enable? I'm happy to upgrade my cluster for fun and it's only running a few things which I can easily port. I feel it'd be useful to define what's required to get it working in 1.14 if it is supported though

Many thanks

[Linutux]

I have had the same issue, the install being stuck. I had nodes with 2 vCPUs and 2GB of RAM for this. I upgraded them to 4 vCPUs and 8GB of RAM. Then the installation went smoothly. Afterwards I downgraded the nodes again.

[davidqhr]

Thanks for the feedback. Sorry, it seems that the document is not clear enough. The installation failure of 1.14 should be mainly because some of the features that kalm depends on are not enabled on your cluster. In your case, it should be CustomResourceWebhookConversion.

1. I will improve the documentation later, suggesting how to configure the apiserver parameters on the lower version of k8s to ensure that the features required by the kalm function are turned on.
2. Remind everyone to try kalm with a suitable machine configuration, for example (2Core, 4G Memory)