You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
We use Argocd to deploy all of our applications, currently we are using a stateful Kafka setup which deploys three kafak pods and three kafka zookeeper pods. Each of tehse in turn creates a PPVC as such:
The problem we run into is that ownership of these resources does not seem to be correct as when you utilize ArgoCD it see's these resources and gives the error:
This resource is not present in the application's source. It will be deleted from Kubernetes if the prune option is enabled during sync
And when ArgoCD resyncs, it tries to delete the PVC's
My understanding is that if you set the ownership of these resources to the Kafka CRD, this would no longer occur as they would be seen of children of the Kafka instance:
If you set the owner reference, the PVC will get automatically deleted when the owner (the Kafka CR) is deleted. This is not desired as it might lead to data loss. For this reason, we do not set it by default - similarly to how StatefulSet does it for example.
If you want, you can tell Strimzi to use the owner reference in the Kafka CR (through the deleteClaim flag in the storage configuration - check the docs for details). But we recommend this only for development and test clusters. For production, you should ideally just configure your Argo to not delete it. There are multiple past issues and discussions on this topic where you find the various ways.
Bug Description
We use Argocd to deploy all of our applications, currently we are using a stateful Kafka setup which deploys three kafak pods and three kafka zookeeper pods. Each of tehse in turn creates a PPVC as such:
The problem we run into is that ownership of these resources does not seem to be correct as when you utilize ArgoCD it see's these resources and gives the error:
And when ArgoCD resyncs, it tries to delete the PVC's
My understanding is that if you set the ownership of these resources to the Kafka CRD, this would no longer occur as they would be seen of children of the Kafka instance:
something similar to:
on each PVC would solve this issue completely
Steps to reproduce
Deploy a stateful kafka cluster using Argocd application
Expected behavior
We should be able to fully utilize this operator with a GitOps pattern
Strimzi version
0.37.0
Kubernetes version
1.29
Installation method
helm
Infrastructure
Scaleway Kapsule
Configuration files and logs
No response
Additional context
No response
The text was updated successfully, but these errors were encountered: