Can’t break the
uptimestreakInside me there are two wolves: Wanting a nice uptime and wanting to constantly update
systemctl restart takes care of 99% of things for me.
Linux is almost like cheating, where is the fun? The sport? The challenge‽
Sometimes it’s not about the destination, but the journey 😌
I’ve been to the madhouse once. I’m not too keen to return.
The journey might take you across scary CLI land in case your GUI broke once again.
Is that still necessary with linux?
Usually only kernel changes if at all, but they mentioned registry keys.
Ya even in an immutable distro, I find some little things get glitchy without a restart every couple days or so
Kernel updates can require it, but even then you can often get away with replacing the specific module.
Is what necessary?
Reboot to install software?
Generally no. There are some parts of your system that you will have to reboot for (like the kernel). But apps? Installing a new service?
No.
Most systems you just install the app you want, and run it.
There are some immutable distros the require things that are installed as part of the base system to only be available after a reboot, but they provide ways to install things without making it a part of the base system. Thus no reboot required.
Both Windows and Linux are able to reload parts of their kernel without needing to reboot. It’s still suggested, but as long as the software doesn’t actually force you to you can just say “reboot later” then just start the program like normal.
Yes. Using kexec.
Though this is irrelevant for majority of users: I’ve never seen it as the default.
That +
WARNING: Use with caution! Kernel crashes, spontaneous reboots, and data loss may occur!
gives me the idea that its likely safer to just do a proper reboot, if your alternative is kernel patching or loading a completely different kernel.
Plus, its likely that not every single bit of firmware running on your devices support live patching. Thus you will be rebooting eventually, unless you are fine with avoiding the updates.
In some cases, yes.
I’ve run into probably half a dozen applications that request a reboot before installation, even if you just rebooted and the launcher was the first thing you ran.
Usually older and very niche software. Business shit from the AS400 era, medical applications (including modern), not generally your standard stuff most people are installing.
Ah, I was thinking that with anything modern I haven’t had that. Kernel needs a reboot, services can be restarted and apps just chug along until you restart them yourself
No
It’s advised to reboot after any kernel updates.
deleted by creator
A modern computer will restart and be up and running again within like 20 seconds. This just gives off i aM vErY sMaRt vibes.
Not if it’s running Windows
Yes, even windows.
Not of you need to wait for ever for an update to finish during shutdown, when you just want to pack up your laptop and finally leave the work site - and afterwards it somehow needs the same time to boot up again, because it’s preparing out finishing something, without ever telling you, what the hell it’s actually doing all the time
And usual reboot only takes like 20s, you’re right there, but not with a windows update blocking a normal shutdown or startup
There is no situation where fucking about like they describe is the better option over rebooting
Well, depends on what you need to restart.
If it’s just a service or two, then it’s pretty quick.
Obviously you need to know, what you’re doing.
And I think that’s pretty hard with windows, because it never tells you, what it’s actually doing during an update.With Linux it’s much more transparent and often the restart of a service is just part of my update routine.
No need to close and save all my open stuff, and reboot. Just restart the few services that got an update and that’s pretty much it.So, really depends on the update, the transparency of it and also the personal insight/skill
Computers actually got slower to boot after DDR5. It was true that ultrafast boot would boot in something like 5 seconds from post, now auto memory timings are harder to get right and so require more iterations to achieve stability.
Doesn’t memory training only happen once when you first boot the machine (or reset bios).
I mean having a lot of ram will take a long time to post, but that’s not unique to DDR5. My server is DDR4 with 64 gigs of ram and with it’s original CPU it took ages for the post to finish.
There are memory trainings that happen automatically every time the machine is booted. The more intense ones happen after a full shut down, but ddr4 was far less noticeable than the intense ddr5 trainings which often even need an additional power cycle to get right. These are not your typical automatic timings applied in the bios, they’re trial and error timings set by the firmware of the motherboard I believe.
Sure, if you’re not multitasking and running uninterruptible or PITA-to-get-going-again processes or are just in a good flow where it’ll take way more than 20 seconds to reopen all the programs you had running and breaks it
Na, nothing enterprise boots that fast. Not even with nvme drives.
I had to update a Chromebook-like machine that was running Windows not to long ago. It was excruciating. The restart progress bar on one update after reboot took ~30 minutes to reach 3%.
Keep in mind that the computer is unusable during this time, and all it takes is one poweroff to brick the machine. Ask me how I know :) . I had to leave it plugged in overnight to finish.
If this comment is referring to Windows reboots after update, I will call it confidently incorrect.
