Routine Server Reboots

At one time, it was a common practice of Unix system administrators to reboot their servers regularly, once a month or at another specified interval. This was done to flush the memory, make sure there was nothing wrong with the server, and give it a fresh start.

Nowadays, it seems that many Linux, BSD, and other Unix system admins have abandoned the practice, but there are still some who still swear by it. Is a routine monthly reboot really necessary, or is it just asking for trouble?

On a Linux dedicated server, there is only one time when the system will require you to reboot: after a kernel update. Even with that required reboot, you do not have to perform it immediately after the update. You can schedule it for a low-traffic hour of the day. This is similar on BSD variants and most other Unix-like operating systems.

Excluding those major updates, these servers can run months, even years without needing to reboot, and they can continue to run without any degradation in quality. Processes will not hold memory during general use, and if they ever do, the system admin can usually stop those processes without needing a reboot.

Are routine reboots needed? Probably not, and there is really no good reason to perform them. Nevertheless, if you insist on rebooting regularly, it will not harm your system.