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're running a Kubernetes (k8s) cluster with some machines where each machine has a different kubelet root directory configured as follows:
Machine 1: /container/k8s
Machine 2: /data/k8s
Machine 3: /var/lib/kubelet
The default CephFS configuration examples typically use a single hostPath like this:
Yaml
hostPath:
path: /var/lib/kubelet/...
However, this doesn't account for our multi-root directory setup. How can we configure ceph-csi (specifically for CephFS) to work seamlessly across all nodes considering their differing kubelet root directories?
Question:
What is the recommended approach to ensure that ceph-csi with CephFS can handle multiple kubelet root directories on a 3-node k8s cluster? Are there any best practices or specific configurations within ceph-csi or Kubernetes to accommodate such a scenario?
The text was updated successfully, but these errors were encountered:
@Bpmm9012 yes it's possible but it complicates things as we need to have many daemonset templates which will use different kubelet path(s) which will have customer taints/tolerations to tell where the pod should go, IMHO we can document it but it will be difficult to maintain this yaml, @Bpmm9012 have you tried some symlink or something similar?
@Madhu-1 In appreciation of your explanation, setting up hard symlink seems like an effective approach. I will try to utilize the mount --bind command to made all different directories to the same directory.
Description:
We're running a Kubernetes (k8s) cluster with some machines where each machine has a different kubelet root directory configured as follows:
Machine 1: /container/k8s
Machine 2: /data/k8s
Machine 3: /var/lib/kubelet
The default CephFS configuration examples typically use a single hostPath like this:
Yaml
hostPath:
path: /var/lib/kubelet/...
However, this doesn't account for our multi-root directory setup. How can we configure ceph-csi (specifically for CephFS) to work seamlessly across all nodes considering their differing kubelet root directories?
Question:
What is the recommended approach to ensure that ceph-csi with CephFS can handle multiple kubelet root directories on a 3-node k8s cluster? Are there any best practices or specific configurations within ceph-csi or Kubernetes to accommodate such a scenario?
The text was updated successfully, but these errors were encountered: