cluster-autoscaler-with-capi/docs/plans-executions/2026-09-29-2315-pv1-2-workers.md

Execution log: lab1 workers on both PVE nodes (template copy per node)

Pairs with docs/plans/2026-09-29-2315-pv1-2-workers.md.

Status

• Step 0 — Record the plan
• Step 1 — vmbr0 on pv1-2 VLAN-aware (owner)
• Step 2 — Template 9101 on pv1-2 (owner)
• Step 3 — Second worker MachineDeployment lab1-workers-pv1-2
• Step 4 — Scale test across both nodes
• Step 5 — Docs and history

Step 0 — Record the plan (2026-09-29 23:15 CEST)

Plan and this log committed on their own, together with the session note that motivated them.

Starting state (read-only PVE API):

• cluster homelab: pv1-1 192.168.0.8, pv1-2 192.168.0.10
• vmbr0 on pv1-2 has no bridge_vlan_aware
• template 101 is only on pv1-1, with disks scsi0 + efidisk0, both on local-lvm
• nextid is 103

Steps 1–2 — bridge and template on pv1-2 (owner, verified 2026-09-30 08:38 CEST)

The owner made vmbr0 VLAN-aware on pv1-2 (/etc/network/interfaces, then ifreload -a):

bridge-vlan-aware yes
bridge-vids 2-10

This deviates from the plan: bridge-vids is 2-10, not 2-4094. That's enough for VLAN 4, and any VLAN above 10 on pv1-2 would need the range widened first.

Template copy, run on pv1-1. A cross-node clone is refused from local-lvm, so the copy is a clone on the same node followed by an offline migration:

qm clone 101 9101 --full 1 --name ubuntu-2404-kube-v1.35.9 --storage local-lvm
qm migrate 9101 pv1-2 --with-local-disks
ssh root@192.168.0.10 qm template 9101

The PVE API shows 9101 on pv1-2 with template: 1, and scsi0/efidisk0 as local-lvm:base-9101-disk-*.

Smoke test (image-builder-proxmox.md) on pv1-2, with TPL=9101 ID=9901: the VM got a VLAN 4 DHCP address through the guest agent, and was destroyed afterwards. Before running it:

• /root/jan.pub had to be copied to pv1-2 first: scp ~/.ssh/id_ed25519.pub root@192.168.0.10:/root/jan.pub.
• The doc's ID=9001 now collides with debian-13-template, and the snippet ends with qm destroy $ID. The snippet now uses 9901 and checks that the ID is free (a9f5879).

Worth noting:

• VMIDs are cluster-wide (pmxcfs), which is why the copy needs its own ID. Names don't have to be unique.
• The copied efidisk0 reports size=528K (the source says 4M). This doesn't matter: the template boots SeaBIOS.