

2.5Gb won’t be your bottleneck on the WAN side — 3-4TB a month is only about 10Mbps averge, and even torrent bursts won’t saturate that unless your internet is over a gig. The place I’d watch is traffic hairpinning through the router, like cameras and media going via OPNsense instead of staying on the switch — keep the NVR/Frigate traffic local where you can.
If you do virtualise it, the security bit that matters most is isolation: keep the WAN side separate from your media containers, and keep the Proxmox managment interface on the LAN only, not reachable from WAN. FYI, that’s also how a misconfigured bridge ends up exposing the host.
And keep your old Archer handy as a fallback router with the OPNsense config backed up — if Proxmox needs a reboot, you can swap one cable and be back online. YMMV, but for a house with 50 devices, I’d rather have that fallback than squeeze everything onto one box.
Yeah, updating is the obvious move, but the annoying part for selfhosters is finding all the places nginx is actually running. It’s not just the main server — it’s hiding inside Nginx Proxy Manager, SWAG, various docker images, and whatever reverse proxy your apps shipped with. I went through mine last night and found two containers I had completely forgotten about still on an old image.
Quick way to check is
nginx -von the host, plusdocker psand looking at image tags, then actually restarting the containers after you pull the new image — pulling alone doesn’t do anything untill you recreate them. YSK that disabling HTTP/2 is a decent temp fix if you’re stuck on an old version for some reason, but honestly just update, check your logs for anything odd, and make a note of where each instance lives so you don’t have to hunt for them again next time you recieve a CVE alert like this.