Hacker News Reader

Comment by spudlyo | Hacker News Reader

Every day. I have a long running tmux session running irssi on a community Linux host that's been humming along in some forgotten corner of a vast data center for the past 11.71 years[0]. My ~/.irssi/config has me on a number of channels on a handful of different servers. Some on networks like chat.freenode.net and irc.mibbit.net, and some on single servers hosted by different groups of friends.

Freenode has been really helpful for me. I've had great conversations on ##aws, #lfs, #tmux, and #hammerspoon in recent weeks.

[0]:

    root@eidolon:~# smartctl --all /dev/sda | grep Power_On_Hours | cut -c88-
    102567

View on HN

##security and ##ibmthinkpad are my personal favorites. They don't take themselves too seriously (as in, you're allowed to get off-topic), but can be super helpful when needed. The wealth of knowledge combined with the general lightheartedness is refreshing.

Waait a minute. I know we used to love sharing uptime on IRC back in the 90s but these days it's just irresponsible to leave a server running for 11 years without reboots.

Because when has it ever been patched? You know vulnerabilities in IRC clients do exist. That's why we were always told to not irc with root back in the days of BitchX.

I think that command is showing you the total number of hours the system has been running, not the uptime - so it could have been rebooted for patching

[Edit] - Slightly pedantically, it's the number of hours one of the disks has been powered up

Yes I suspected as much. So really this is bragging about continuous power for 11 years. Which is impressive.

I never saw anyone measure uptime in that way before. But it make a lot more sense than measuring the OS uptime.

That's right. We've got 4 bog standard 750G WD Caviar Blues in a software RAID array in that machine, all powered up for 11+ years, none of which appear to have caught any errors during that time. Pretty amazing.

The OS has been updated and the box rebooted for kernel updates a number of times over the years.

I've never used them, but there does appear to be solutions such as kpatch, to patch a kernel whilst running.

That's only intended to be a band-aid until you can do a reboot without heavily impacting users. Say your monthly maintenance isn't for another two weeks, but there's a kernel vulnerability that's very worrisome. You may be able to kpatch it, but you do still need to do a reboot later.

Reply on news.ycombinator.com