|
|
|
|
|
by SirMaster
28 days ago
|
|
Is it weird that we don't have a way to update the system without rebooting by this day and age? I think there are things in Linux like live kernel patching (kexec, ksplice), but why by 2026 is this sort of architecture of live system updates not a common or included feature yet? People hate updates and having to reboot and have downtime right? Security updates to core systems are more important than ever now too. So why hasn't this problem been tackled I wonder? Wouldn't it be great if our systems could update without us having to reboot and interrupt our workloads? |
|
The complex monsters tend to result in corner case considerations like "already open sockets cannot be upgraded but new ones will be" and a massive amount of testing to make sure the convenience saved is more than the risk of inconvenience due to some bug with a complex live patch.
Designing the whole system for live patching from day 1 can reduce the number of complex monsters but it's still more complex, and that's a ton of work (rebuild all or just shipping live patches) that could go into fixing other annoyances/limitations which are even mor inconvenient than rebooting for most users.
The result is some things support them in the most impactful cases, like certain types ofOS kernel updates, but most cases it's better to work on other conveniences instead.