Closed psamagal closed 3 years ago
Did you used the Helm2 Chart from https://github.com/strimzi/strimzi-kafka-operator/releases/tag/0.20.0? Or just the Helm3 Chart from the index on the website? The Helm3 Chart has the CRDs in a subfolder or something like that so maybe Helm2 did not installed them?
(the error from the Cluster Operator basically says that the CRDs are not installed)
Yes, i used the v0.2.0 helm release version. You can check my only code on the terraform snippet above.
What else should i do to install the CRD's? is there a separate deployment? Or do i need to go with Helm3?
My understanding is that with Helm you should use the strimzi-kafka-operator-helm-2-chart-0.20.0.tgz chart. But I do not know much about Helm TBH within or without Terraform. But the idea definitely is that it installs the CRDs as well.
That was what i was expecting, the CRD to be installed, and it worked well with v0.19.0 of the chart. I will try to enable debug on terraform helm provider to see if something pops up, during the apply. If not, I will try a manual install of the operator to see, if the issue persists.
So back to my original question which I'm still not sure you answered. Did you used the Helm2 Chart from the release page or not? Because the index now links tot eh Helm 3 Chart as Helm2 is end-of-life.
No, i did not download it from release page. i did only the terraform above, any download was made by terraform itself, will try to run in debug to find out, what terraform does in the background. This is the helm provider configuration
provider "kubernetes" {
host = var.gke_cluster_api
token = var.gke_cluster_token
cluster_ca_certificate = var.gke_cluster_ca_certificate
load_config_file = false
}
provider "helm" {
version = "0.10.2"
install_tiller = false
kubernetes {
host = var.gke_cluster_api
token = var.gke_cluster_token
cluster_ca_certificate = var.gke_cluster_ca_certificate
config_context = "none"
}
}
Yes, is possible a good idea to start playing with helm 3. :D
A quick test with helm 3 + minikube and most recent terraform provider for helm. it did work well. With Helm 2 + minikube + old terraform provider (max version, before helm3 was enforced) it was failing.
I will just update to Helm 3, luckily new version does not need tiller anymore, so i can upgrade projects 1 by 1. Will close this, don't believe it's worth it to chase the issue around, on old versions.
Thanks @scholzj for the help.
I did try to do a deploy to a K8S cluster (v1.15) of the latest strimzi kafka operator using the Helm (v2) Chart. The version of the chart is strimzi/strimzi-kafka-operator 0.20.0
But when strimzi-cluster-operator pod starts i obtain the following error, no other config changes are being applied apart from default values of the chart
If i tried to deploy the previous version (0.19.0) of the helm chart, all works fine and the operator is deployed with success. I had dig a little on the differences between what is deployed between 0.19.0 and 0.20.0 and i notice that it appears that with v0.20.0 the CRD are not deployed.
Looking to the helm code i see, that is a property createGlobalResources that enable or not the deploy, but by default is value is already true.
I'am deploying the operator with terraform helm provider. (v0.12.2)
So i get no idea, why they are not being deployed, causing the above issue.
Did i miss any value? or i need to do anything else on version 0.20.0? Any help would be appreciated. Thanks