Longhorn
This repository uses Longhorn, delivering simplified, cloud-native persistent block storage without the cost overhead of open core or proprietary alternatives.
Node Drain
The node drain behavior is controlled with longhorn_vars.kubernetes.default_settings.node_drain_policy and longhorn_vars.kubernetes.default_settings.pod_deletion_policy variables, defined into main.yaml defaults file.
Example of a manual node drain:
$ kubectl drain hypnos --delete-emptydir-data --ignore-daemonsets
node/hypnos cordoned
Warning: ignoring DaemonSet-managed Pods: kube-system/cilium-envoy-dksxd, kube-system/cilium-rhhlf, kube-system/engine-image-ei-51cc7b9c-zbx65, kube-system/kured-gxkjq, kube-system/longhorn-csi-plugin-xv5xz, kube-system/longhorn-manager-zr7qs, kube-system/vls-vector-wsmbq, kube-system/vmks-prometheus-node-exporter-7z4zx
evicting pod kube-system/argo-cd-argocd-application-controller-7c768c5fb8-vspvg
evicting pod kube-system/argo-cd-argocd-repo-server-54bd46d95b-zrr45
evicting pod kube-system/instance-manager-d2ee416cd23fb2962e3ddee73dd8b374
evicting pod kube-system/vmks-victoria-metrics-operator-79588f8bb7-bzp4c
evicting pod kube-system/vmsingle-vmks-victoria-metrics-k8s-stack-7dc6b865fb-n8lhb
pod/argo-cd-argocd-application-controller-7c768c5fb8-vspvg evicted
pod/argo-cd-argocd-repo-server-54bd46d95b-zrr45 evicted
pod/vmsingle-vmks-victoria-metrics-k8s-stack-7dc6b865fb-n8lhb evicted
pod/vmks-victoria-metrics-operator-79588f8bb7-bzp4c evicted
evicting pod kube-system/instance-manager-d2ee416cd23fb2962e3ddee73dd8b374
error when evicting pods/"instance-manager-d2ee416cd23fb2962e3ddee73dd8b374" -n "kube-system" (will retry after 5s): Cannot evict pod as it would violate the pod's disruption budget.
evicting pod kube-system/instance-manager-d2ee416cd23fb2962e3ddee73dd8b374
error when evicting pods/"instance-manager-d2ee416cd23fb2962e3ddee73dd8b374" -n "kube-system" (will retry after 5s): Cannot evict pod as it would violate the pod's disruption budget.
evicting pod kube-system/instance-manager-d2ee416cd23fb2962e3ddee73dd8b374
pod/instance-manager-d2ee416cd23fb2962e3ddee73dd8b374 evicted
node/hypnos drainedDuring the node drain process, monitor the node into Longhorn UI:
And the related volume replicas rebuild status:
