In this video I’m continuing my Kubernetes homelab journey, and we’re finally fixing one of the most annoying parts: storage. When you’re running a barebones cluster (like mine) you end up manually provisioning volumes, fighting mount paths, permissions, and user IDs—especially when you’re deploying apps like Uptime Kuma. So I’m switching to Longhorn, which auto-provisions persistent volumes whenever you create a PVC, and gives you a clean UI so you’re not constantly correlating PVs and PVCs in kubectl. I walk through the setup end-to-end: installing the iSCSI initiator utilities on every node, deploying Longhorn with Helm into a dedicated namespace, and then exposing the Longhorn frontend through MetalLB by patching the service to a LoadBalancer with a specific IP. I also show the important homelab-specific bits: updating both replica settings (defaultReplicaCount and defaultClassReplicaCount) so you don’t accidentally request more replicas than your node count can support, and adding dedicated disks per node (I used LVM-backed mounts) instead of letting Longhorn fill up your root filesystem. To prove it works, I redeploy Uptime Kuma and let Longhorn dynamically create the PVC/PV—plus I cover a webhook/backup target hiccup I hit and how I worked around it.