diff --git a/02-k8s-bootstrap/CLAUDE.md b/02-k8s-bootstrap/CLAUDE.md
index f360f92..fc7d1cf 100644
--- a/02-k8s-bootstrap/CLAUDE.md
+++ b/02-k8s-bootstrap/CLAUDE.md
@@ -11,7 +11,8 @@ Bootstraps a k3s cluster on the Proxmox VMs created by `proxmox-infra`. Starts V
 3. Installs k3s on `k3s-master-1` with `--cluster-init --tls-san <master1Ip>`
 4. Joins `k3s-master-2` and `k3s-master-3` as embedded etcd nodes
 5. Joins `k3s-worker-1` and `k3s-worker-2` as agent nodes
-6. Reads `/etc/rancher/k3s/k3s.yaml` from master-1 via SSH, patches the server URL, and exports it as the secret stack output `kubeconfig`
+6. Formats the workers' dedicated `scsi1` disk (ext4, if not already formatted) and mounts it at `/mnt/longhorn-extra` (via `/etc/fstab`) — this is the disk `03-k8s-infra` later registers as an extra Longhorn disk, to keep Longhorn off the root filesystem and avoid `DiskPressure`
+7. Reads `/etc/rancher/k3s/k3s.yaml` from master-1 via SSH, patches the server URL, and exports it as the secret stack output `kubeconfig`
 
 VM IDs and the CI runner SSH private key are read automatically from the `proxmox-infra` stack output via StackReference — no manual setup needed for those.
 
diff --git a/03-k8s-infra/CLAUDE.md b/03-k8s-infra/CLAUDE.md
index 0b0edfd..0c0235d 100644
--- a/03-k8s-infra/CLAUDE.md
+++ b/03-k8s-infra/CLAUDE.md
@@ -21,7 +21,7 @@ CI installs the `helm` CLI (`azure/setup-helm`) before running Pulumi.
 | `truenas-nfs` StorageClass | Backed by an NFS share on TrueNAS (`Retain` reclaim policy) |
 | cert-manager | Certificate lifecycle management |
 | `letsencrypt-prod` ClusterIssuer | DNS-01 via Cloudflare, issues Let's Encrypt TLS certs |
-| Longhorn | Default block storage (`longhorn` StorageClass, 3 replicas) — used by app/monitoring PVCs |
+| Longhorn | Default block storage (`longhorn` StorageClass, 2 replicas) — used by app/monitoring PVCs. Worker nodes get an extra `longhorn-extra` disk at `/mnt/longhorn-extra` (a dedicated mount set up by `k8s-bootstrap`, patched onto the Longhorn `Node` CR here) so Longhorn data stays off the root disk and doesn't cause `DiskPressure`. `nodeDownPodDeletionPolicy` deletes both the StatefulSet and Deployment pod on node-down so stateful workloads reschedule automatically |
 | kube-vip | DaemonSet that ARP-announces a virtual IP for LoadBalancer Services |
 | kube-vip-cloud-provider | Allocates LoadBalancer IPs from a single-address pool (`kubeVipAddress`) — k3s's built-in ServiceLB is not used |
 | `traefik` ServicePatch | Pins Traefik's LoadBalancer IP to `kubeVipAddress` |