Today I decommissioned an old server, that had an uptime of 4130 days.

Nice wtf

@Jifuna
It was never ever rebooted, or ever patched lol

@Linux What did you run on it? :p

@Jifuna
Not me, a conpany that I helped moved their datacenter today.

It was just a fax server

Did you keep any records of the versions of the software it was running? Could very well be Linux 2.6 etc with such a ridiculous uptime :')

@Condor he confirmed it probably is 2.x by saying it's never been patched, but technically you can upgrade the kernel without ever going down.

I've seen servers with uptimes in the thousands that were fully patched, running the latest kernel and all.

@deadPix3l personally I'd choose running multiple redundant servers and rebooting each one about once a month any day, because it offers redundancy and a reboot is a far more stable application of kernel patches than Livepatch or kpatch can be. But yeah it's *technically* possible.. just that I find redundancy and reboots to be more elegant, especially considering that the old code blocks are not purged during a kpatch afaik, instead the new code is just inserted and new pointers to the newly inserted code are created. So yeah, no kpatch here. The alternative of multiple redundant servers does add to the complexity and cost though, so there's that...

Somewhere there is a server everyone forgot about that has been running since the 80s

@kenogo it is, but that doesn't prohibit changing code at runtime

@kenogo how it works at a technical level is complicated and I won't pretend to understand it but from my understanding:

The kernel has a special kmod which allows functions to be replaced on the fly. The old function still sticks around in slack memory and probably won't go away for a loooooong time, thus some people don't like live patching. But the kernel acknowledges the updated version and it works.

I'm only aware of this in an academic manner. I've seen servers that it's been done to, but I've never personally done it. I just reboot for upgrades.

I think the most fascinating thing here is that there was no power outage in 11 years there!

@hidingFromBoss
Several UPS and Diesel generators.
A real DC m8

@Condor
I am pretty sure it was 2.6 yes :)
I remember when 2.6 arrived, it was a HUGR boost between 2.4 and 2.6

A moment of silence

So you slaughter servers.
RIP server.
🕯🕯🕯

Rest in peace <3

It's still surprising that a machine sustained for that long without any failure causing any reboot

Rip

@Condor you find reboots to be elegant?
Who are you, president of the Windows fan club?

@nathanchere I like stable, up-to-date kernels over dirty patches, or perhaps even none at all. Also I hate Windows with a burning passion. Not like that craptacular OS can gather more than a few days of uptime without crashing anyway. But a scheduled reboot a month, while another server's still around to accept new connections? Perfectly good in my book.

Out of curiosity, how'd you go about ensuring high availability? How do you maintain your servers, if any?

@Condor

There is alot of ways to have HA on Windows, infact - many servers does have better HA on Windows than Linux.

Respect. 11.35 years without a single reeboot. Damn man that server is 3 years older than my younger brother.

@Admin-who did you reboot your brother, though?

@electrineer and how do you do that lol.

@electrineer running Novell or NT too.