Andrew Mercer
on this page

Key settings (kubectl get settings.harvesterhci.io)

Setting Purpose
backup-target S3/NFS target for VM backups
storage-network Dedicated Longhorn replication network
overcommit-config CPU/memory/storage overcommit ratios
vm-force-reset-policy Restart VMs from failed nodes after timeout
cluster-registration-url Import into Rancher
ssl-certificates UI/API TLS certificate
additional-ca Trust internal CAs (image downloads, backup targets)
http-proxy Egress proxy configuration
ntp-servers Time sync
auto-disk-provision-paths Auto-add matching disks to Longhorn
containerd-registry Registry mirrors/credentials
upgrade-checker-enabled Online version checks
csi-driver-config Third-party CSI driver mapping
default-vm-termination-grace-period-seconds Shutdown grace for VMs
support-bundle-* Support bundle timeouts/images

Namespaces to know

Namespace Contents
harvester-system Harvester controllers, webhooks, KubeVirt, upgrades
longhorn-system Longhorn
cattle-system Embedded Rancher, cluster-repo, agents
kube-system RKE2 system components, kube-vip, Multus
harvester-public Shared images and templates
cattle-monitoring-system / cattle-logging-system Observability add-ons

Default ports

Port Use
443 Harvester UI/API on the VIP
6443 Kubernetes API
9345 RKE2 supervisor (node join)
2379–2380 etcd
10250 kubelet
8472/UDP VXLAN (Canal)
22 SSH (rancher user)

← Gotchas and hard-won lessons · Index