Proxmox

Sooner or later almost every Proxmox host grows a USB device that needs to talk to one specific guest. A Zigbee or Z-Wave radio for Home Assistant, a TV tuner for a media server, a serial adapter for a UPS, a security dongle for some old piece of software that still checks for one. Getting that device into the right VM or container is usually simple the first time and then mysteriously stops working after the next reboot, because the setup that worked was tied to something that wasn’t actually stable. Here’s how passthrough actually works for both VMs and LXCs, and where the stability problem comes from. ...

Every Proxmox host ships with snapshots built in, and they’re so easy to take that it’s tempting to treat them as your backup strategy. Click “Snapshot” before a risky change, click “Rollback” if it goes wrong, done. That workflow is genuinely useful, and you should use it. It is also not a backup, and conflating the two is one of the more common ways homelab operators discover, at the worst possible moment, that they had less protection than they thought. ...

Most homelab buying guides start with the case or the drives, but the CPU and motherboard decide almost everything downstream of them. PCIe lane count sets how many cards you can run at once. ECC support decides whether a bit flip in RAM silently corrupts your ZFS pool or gets caught and corrected. IOMMU group layout decides whether GPU or HBA passthrough to a VM is a ten-minute config change or a wall you can’t get past no matter what you try. None of that shows up on a spec sheet the way clock speed and core count do, which is exactly why it gets skipped over until someone hits it the hard way. ...

Proxmox ships with a real, iptables-backed firewall built into the platform, and most homelabs never turn it on. It’s not enabled by default at any level, so unless you’ve gone looking for it in the UI, there’s a good chance your VMs and containers are sitting behind nothing but whatever VLAN segmentation you’ve set up, if any. That’s often fine. It stops being fine the moment you’re running something internet-facing, or you want one VM to be reachable only from your management network and nowhere else, or a compromised guest shouldn’t be able to freely talk to everything else on its VLAN just because they share a broadcast domain. This is what the built-in firewall is actually for, and it’s worth understanding the shape of it before you flip it on. ...

The Proxmox-vs-ESXi comparison on this site already covers why so many homelabbers are moving off VMware since the Broadcom licensing changes gutted the free ESXi tier. This one skips the “should I switch” question and gets into the part that actually trips people up: what breaks when you move a real VM from ESXi to Proxmox, and how to get ahead of it instead of debugging a guest that won’t boot at 11pm. ...

A UPS without a way to talk to your hosts is just an expensive surge protector with a battery in it. It’ll keep the lights on for a few minutes, but the moment that battery runs out, every VM and container you’re running goes down exactly as hard as it would have without a UPS at all, just a little later. The part that actually protects you, a UPS that tells your hosts “power’s out, shut down now while you still can,” is a separate piece of software most people never get around to setting up. Here’s how to size the UPS itself, and how to wire up Network UPS Tools (NUT) so an outage ends in a clean shutdown instead of a dirty one. ...

Every homelab operator hits the same wall eventually: you need a fifth Ubuntu VM for some new service, and the thought of walking through another OS installer, setting the hostname, adding your SSH key, and running updates makes you want to just leave the last one running forever instead. Cloud-init templates fix this. Build the template once, and every new VM after that is a clone plus a few lines of config, done in under a minute instead of fifteen. This is the same workflow cloud providers use to spin up instances on demand, and Proxmox supports it natively, no extra tooling required to get the basics working. ...

Most homelab hardware guides stop at “buy it, rack it, install Proxmox.” The thing that actually changes how a homelab feels to run day to day is whether you can reach the machine when the OS won’t boot, the network stack is misconfigured, or the box is three rooms away in a closet with no monitor attached. That’s what out-of-band management is for, and if you’re running used enterprise gear, you almost certainly already own it. Most people never turn it on. ...

Ceph shows up in every “how do I get real shared storage for my Proxmox cluster” conversation, and for good reason, it’s the one option that’s fully native, fully integrated into the Proxmox web UI, and genuinely production-grade software running underneath some very large real-world deployments. It’s also the option most likely to leave a home cluster worse off than before you added it, because Ceph’s design assumptions (plenty of nodes, plenty of network, plenty of dedicated drives) don’t bend easily to a 3-node home lab built from whatever hardware was on hand. Here’s what Ceph actually is inside Proxmox, what it needs to run well, and when it’s the right call instead of the simpler alternatives. ...

Every data hoarder eventually hits the same wall: the motherboard only has six SATA ports, and the build calls for twelve drives. The instinctive fix is a cheap RAID card off eBay, and that instinct is almost always wrong. The actual answer, in nearly every homelab storage build, is an HBA flashed to IT mode. Here’s what that means, why it matters, and how to do it without turning a $40 card into a paperweight. ...