Important bit of #linux advice: Even though a linux system can manage hundreds of days of uptime you should reboot it on a regular basis anyway.

The reason is simple: 100 days of uptime means you haven’t tested the boot sequence for over 3 months.

  • Is the bootloader still working?
  • do you still know if you started a new service but forgot to add it to the boot squence?
  • Is that one service that has its configuration in ram saved?
  • [Edit: added] Bug/security fixes will only become active when programs are restarted and a reboot is the easiest way t make sure everything is the latest version installed.

If you reboot regularly you notice those mistakes while you still remember making them, if the box is forced to reboot when you least need it half a year later you have no idea what’s not working and no time to debug.

@slatian also the torvalds program does NOT like it if you don't reboot after a kernel upgrade

@slatian

That would only apply if you start new services or make significant changes to the system

I have a Raspberry Pi 2 with an uptime of 1733 days and I don't plan to restart it

@sibrosan

I was at this point with ~400 days of uptime, then the pi rebooted because of a power outage and I discovered that the SD-card had sectors that were long gone, I had the same problem again a year later, because one doesn’t believe that this is a problem that can happen multiple times the first time it happens.

@slatian

My Pi does not write to its SD card a lot and it is protected by battery backup against power loss while writing. So i think the risk of SD card corruption is rather low.