Server maintenance

[on:]

We don't seem to have a thread to announce server maintenance instances, so it will go here.

Sometime during Saturday 2 December CET, the server will be down for a reboot.

$ uptime
 08:05:56 up 1000 days, 20:53,  4 users,  load average: 0.86, 0.80, 0.79

While such uptime is enough to make any Linux user proud, I also think it means it is time for a reboot, so it can get on a much newer kernel.

I cannot tell you how long it will last, hopefully just half an hour, but depending on how it goes, it might take longer.  I will post in here once the job is done.  But don't be surprised if PEEL isn't responding on Saturday.

[Reply #1 on:]

I am reserving this post for a future, where I might have several of these announcements.

[Reply #2 on:]

Thanks for the heads-up, Svip. The server I set up at work has only been up for about a year. 1000 days is great. Good luck with the update!

[Reply #3 on:]

Looks like the reboot went without a hitch.

The stats before I rebooted:

 10:28:33 up 1003 days, 23:16,  1 user,  load average: 1.27, 0.96, 0.81
 Linux saxony 4.19.0-14-amd64 #1 SMP Debian 4.19.171-2 (2021-01-30) x86_64 GNU/Linux

And now:

 10:43:26 up 10 min,  3 users,  load average: 0.71, 1.80, 1.55
Linux saxony 5.10.0-26-amd64 #1 SMP Debian 5.10.197-1 (2023-09-29) x86_64 GNU/Linux

Let me know if you spot any oddities.

[Reply #4 on:]

We reboot the instrument servers in my lab once every few weeks just to make sure that they don't develop any quirks*. Kinda like wiping a droid's memory every six months in the Star Wars universe.

So from my perspective, a thousand days of continuous uptime is actually mildly horrifying.

*It's less about preventing them from becoming sentient and overthrowing humanity, and more about making sure that they handshake with the network properly in the wake of any updates.