Yup. My entire PC desk (monitor, PC, 2+1 speakers) draw 7W when the PC is turned off (old speakers draw power when off for some reason). For comparison: My NUC server draws 7W white turned on, doing useful work. This infuriates me, so I got a zigbee power switch and shut the PC desk completely off when I’m not home.
If 7W for nothing pisses me off, you’re damn straight an idle or sleeping PC will too!
On the other hand, if the computer is powered on, there’s wear and tear on the moving parts (mostly just fans now), and components like capacitors have a limited lifespan. These tend to be the first components that fail anyway, and I’ve always thought it odd to further reduce the lifespan of these components with the hope to extend the lifespan of what’s already the most reliable parts of the computer.
Now, with modern computers that sip power at idle but can consume hundreds of watts under load, the difference in temperature at idle and load is much greater than room temperature (off) and idle, so even if I was worried about thermal cycling I’d still be inclined to turn the computer off when it’s not needed because when it’s off there will be no big temperature swings. Granted, with Linux when my PC isn’t being used it pretty much just sits at a constant and steady idle… but Windows on the other hand…
Combined with the added electricity cost of not running the computer when it’s not needed, not leaving the machine running all the time is the obvious choice.
You’ve got to reboot after kernel updates, otherwise it can’t load new modules. I’ve been confused at least twice why something didn’t load until.I remembered the reboot.
I think the command “systemctl kexec” would like to have a word. Great command to know if you have a VM on a system you dont manage / share with others.
So you can just run kexec if its installed on the distro. This tells the kernel to boot into another kernel. The reason to use it with systemctl is to properly shut down all the services running in userspace. That command will have systemd gracefully turn off all services and then the new kernel with whatever updates / modules can be loaded in a clean environment.
Its useful if say, you have a VM in a data center. Now most of them provide a web gui where you can turn your VM off and then on. But if you’re lazy like me and already remoted into the terminal lol
but I still don’t get it. couldn’t you just systemctl reboot? it will boot the new kernel that way too. its supposed to be the same as from the web gui
If you run systemctl reboot on a non-vm it will actually power cycle the system and cause it to go back through the BIOS and then the bootloader. Using systemctl kexec allows you to “restart” the computer without having to go all the way back through the full boot process.
In the case of a VM, some are setup to do this behind the scenes. For example, virt-manager allows for direct kernel booting. If you look in the options there will be a path to the kernel. If its not setup that way, then the VM still has a bootloader. In that case, restarting the VM with kexec will allow for a faster reboot since the bootloader is skipped completely.
What’s this “boot” of which you speak?
Do people really turn their machines off these days?
Yup. My entire PC desk (monitor, PC, 2+1 speakers) draw 7W when the PC is turned off (old speakers draw power when off for some reason). For comparison: My NUC server draws 7W white turned on, doing useful work. This infuriates me, so I got a zigbee power switch and shut the PC desk completely off when I’m not home.
If 7W for nothing pisses me off, you’re damn straight an idle or sleeping PC will too!
Yes, I’m not wasting my hardware life and electricity for no gain.
Its arguable that machines last longer staying at temp.
I use my machines enough that having to boot when I need it is just time wasting.
Thermal cycling
On the other hand, if the computer is powered on, there’s wear and tear on the moving parts (mostly just fans now), and components like capacitors have a limited lifespan. These tend to be the first components that fail anyway, and I’ve always thought it odd to further reduce the lifespan of these components with the hope to extend the lifespan of what’s already the most reliable parts of the computer.
Now, with modern computers that sip power at idle but can consume hundreds of watts under load, the difference in temperature at idle and load is much greater than room temperature (off) and idle, so even if I was worried about thermal cycling I’d still be inclined to turn the computer off when it’s not needed because when it’s off there will be no big temperature swings. Granted, with Linux when my PC isn’t being used it pretty much just sits at a constant and steady idle… but Windows on the other hand…
Combined with the added electricity cost of not running the computer when it’s not needed, not leaving the machine running all the time is the obvious choice.
TIL, tyvm!
You’ve got to reboot after kernel updates, otherwise it can’t load new modules. I’ve been confused at least twice why something didn’t load until.I remembered the reboot.
I think the command “systemctl kexec” would like to have a word. Great command to know if you have a VM on a system you dont manage / share with others.
won’t this kill all their processes just the same?
So you can just run kexec if its installed on the distro. This tells the kernel to boot into another kernel. The reason to use it with systemctl is to properly shut down all the services running in userspace. That command will have systemd gracefully turn off all services and then the new kernel with whatever updates / modules can be loaded in a clean environment.
Its useful if say, you have a VM in a data center. Now most of them provide a web gui where you can turn your VM off and then on. But if you’re lazy like me and already remoted into the terminal lol
but I still don’t get it. couldn’t you just systemctl reboot? it will boot the new kernel that way too. its supposed to be the same as from the web gui
If you run systemctl reboot on a non-vm it will actually power cycle the system and cause it to go back through the BIOS and then the bootloader. Using systemctl kexec allows you to “restart” the computer without having to go all the way back through the full boot process.
In the case of a VM, some are setup to do this behind the scenes. For example, virt-manager allows for direct kernel booting. If you look in the options there will be a path to the kernel. If its not setup that way, then the VM still has a bootloader. In that case, restarting the VM with kexec will allow for a faster reboot since the bootloader is skipped completely.
oh, so its useful for VMs that do direct kernel booting! will a simple reboot just fail on those systems?
I didn’t know about that one.
I hope it helps you someday! :)
Gtk what replaced ‘telinit 3’
yes
Lol, have my upvote
Every day