The Illusion Of Uptime
For many years a common bragging point for Linux based users was uptime. Browse the web to find many boasts about uptime and not rebooting a server.
For security aware users that day no longer exists and hasn’t for many years. With many distros kernel updates now are an almost weekly affair.
While real-time updating of the kernel exists, few people are using this feature.
In my daily experience I notice that uptime is meaningless with respect to troubleshooting and debugging systems. Once upon a time users could brag about restarting services and daemons rather than reboot. I do this often too, but my experience is that reboots often are useful to fix buggy behaviors. Just clear the slate.
A challenge with these weekly kernel updates is rebooting important infrastructure servers. I do this at work. We reboot servers at lunch or late in the evening. A part I dislike is I reboot remotely and rack servers take a much longer time to reboot than a consumer grade motherboard. Much longer, thanks to RAID controllers. Much like a spaceship reentry into the atmosphere when radios blackout, these reboots leave me waiting for the duration. I always somewhat sigh in relief when I can again SSH into a server after a reboot.
I am curious how experienced sysadmins handle this now regular routine of rebooting critical servers.