Published Nov 20, 2026, 8:01 PM EST His love of PCs and their components was born out of trying to squeeze every ounce of performance out of the family computer. Tinkering with his own build at age 10 turned into building PCs for friends and family, fostering a passion that would ultimately take shape as a career path. Besides being the first call for tech support for those close to him, Ty is a computer science student, with his focus being cloud computing and networking. He also competed in semi-pro Counter-Strike for 8 years, making him intimately familiar with everything to do with peripherals. Proxmox is a hypervisor built for always-on servers that never get a break from duty, and because of that, the defaults set when you install it on your home lab aren't necessarily the most optimal. On a fresh install it can hold your CPU at higher clocks than necessary and waste disk space on the SSD you boot from. For a home server or lab that idles the vast majority of the time that it's on, these small defaults can add up, but thankfully, saner settings are relatively easy to configure. Proxmox will run your CPU like it's always under load It'll keep the clocks higher than they need to be set Potentially the biggest waste of power on your Proxmox box is the CPU. By default, the CPU governor is set to the performance mode, which isn't a problem on the surface, but for a home box that's not doing much 99% of the time, it's a big waste. The CPU governor is the Linux kernel setting that decides how high your CPU clocks itself, and setting it to performance tells it to stay near its maximum frequency regardless of the load. To see what yours is set at, you can use the command: cat /sys/devices/system/cpu/cpu0/cpufreq/scaling_governors. What you switch to depends on what CPU you have and which driver is active. Most modern AMD and Intel chips one would have in a Proxmox box would be running intel_pstate or amd-pstate-epp and only offer performance and powersave options. In the powersave mode, your clocks will still ramp up under load, but if your driver is listed as acpi-cpufreq, powersave will lock the CPU at its lowest frequency regardless of load, which isn't what you want either. schedutil is what you would want instead in that case. Check which driver you have with the command: cat /sys/devices/system/cpu/cpu0/cpufreq/scaling_driver before making any changes. There's no GUI toggle for this, but there is a community script available called PVE CPU Scaling Governor that shows your current governor and switches it, along with adding a cron entry so it's not wiped with a reboot. To do it manually, you can take a governor for a spin first before making it permanent with the command echo powersave | tee /sys/devices/system/cpu/cpu*/cpufreq/scaling_governor, where powersave is the governor of choice. To make it permanent, you add cpufreq.default_governor=powersave to GRUB_CMDLINE_LINUX_DEFAULT in /etc/default/grub and run update-grub. Hosts that boot with systemd-boot will need the same parameter put in /etc/kernel/cmdline, followed by proxmox-boot-tool refresh. Some guest operating systems can cope poorly with frequency changes, so if you find your VMs acting up after the change, I'd swap it back as a troubleshooting measure first. ASPM is what allows for deeper idle Another small tweak that can make a difference Editing the CPU governor only covers the cores and their clock speeds, but it doesn't cover the other devices in your system. Part of getting the rest of your system into deeper idle states is PCIe Active State Power Management (ASPM). ASPM can be checked with lspci -vv | grep ASPM, and you'll be looking for mentions of LnkCtl: ASPM Enabled. LnkCap: will show what the link supports, so if you see anything like ASPM not supported, that means the device cannot be configured with ASPM. Toggling ASPM is done through the BIOS, so after you turn it on and reboot, rerun the command to check if it took effect. One caveat: Intel’s I225 and I226 2.5GbE controllers—exceedingly common in mini PCs home labbers use—have a documented habit of dropping off the network when ASPM is enabled, so make sure you can reach the host’s console before you flip it on. Storage thinning Proxmox's defaults here don't make sense for a home user Every VM you create in Proxmox by default has the "Discard" setting unchecked. The Discard setting controls whether the hypervisor tells the bare metal storage if certain blocks are free or not, and it ends up having a pretty big impact on your system's storage. When a guest deletes a file, its filesystem doesn't really erase that thing, it just marks those blocks as free in its own records. This is where TRIM comes in: the guest will later send a list of blocks it no longer needs, either on a schedule through fstrim, or immediately if Discard is checked in the Proxmox GUI. With Discard off, QEMU drops the guest's TRIM requests. The guest doesn't know any better, and it thinks it trimmed properly. With it on, QEMU turns each request into a real deallocation operation on the backing storage. Without Discard enabled, a thin-provisioned disk only ever grows. This is a problem in terms of capacity, but it also affects drive endurance. On a default install, the LVM thin pool passes discards through to the physical drive. You can confirm this with lvs -o+discards pve/data, which should report passdown. A TRIM that leaves the guest can then travel all the way to the SSD’s controller, which marks those flash pages invalid. An SSD can’t overwrite data in place, though: before it erases a block, it has to copy every page it still thinks is live somewhere else. When it knows which pages are dead, garbage collection has less work to do in copying, and that means fewer internal writes, saving some drive health in the process. With Discard off, that information stops at the hypervisor, and the drive keeps carefully preserving data the guest deleted months ago. The potential downside of enabling Discard can come in the form of high performance overhead when performing live migrations, which isn't something you'll often be doing as a home lab user. You can comfortably enable it on your VMs. Proxmox defaults are for the enterprise, not the home user They're an excellent fit for large clusters that are in charge of a large number of users, but Proxmox's defaults are often made without considering the home user at all. One short session of tweaking can bring your home lab more in line with how you actually want it to run, which should be cool, quiet, and not hammering the solid-state storage that is only getting more expensive.
Proxmox ships with settings that waste power and disk space, and one cleanup fixes both
Full Article
Original Source
Read the full article at Xda-developers →KhanList aggregates and links to publicly available news content. We do not host full articles from third-party sources. Always verify important information with original sources.