jellyfin-proxmox-deploymentlisted

# Jellyfin on Proxmox — LXC deployment with Intel QuickSync passthrough Full recipe for deploying Jellyfin as a native (non-Docker) service in an unprivileged Debian 13 LXC on a Proxmox host with a single shared Intel iGPU, exposed only over Tailscale Serve. First built 2026-07-16 in a homelab with one Proxmox host (Beelink N100, Alder Lake-N iGPU) and a Terraform+Ansible-managed fleet; the mechanics below generalize to any single-iGPU Proxmox host. Decommissioned 2026-07-18 at Will's request (a deliberate choice, not a resource-pressure one) — this skill exists so a future redeploy doesn't have to rediscover any of the following. **⚠️ The iGPU is currently claimed by something else.** The same day Jellyfin was decommissioned, its freed iGPU was passed through **exclusively** to the Immich VM via `vfio-pci` (full PCI passthrough — this hardware has no GVT-g/mdev support, so it can't be split). A future Jellyfin LXC redeploy **cannot** get `/dev/dri` back without first reverting that passthrough (see the homelab CLAUDE.md item 17 entry, "iGPU passthrough added 2026-07-18", for the exact revert steps) — check whether the GPU is still exclusively bound to a VM before assuming the simple LXC bind-mount path below will just work. ## The core architectural decision: LXC, not VM, despite the general rule This homelab's standing rule is "GPU passthrough need → VM" (hardware passthrough generally wants a VM for clean device isolation). **Deliberately override that rule for Jellyfi