Hmm. I’m in the same boat as Warren. Laptop gets turned off every night (only takes a few seconds to boot, so it makes sense) and no desktop. But my server has been running for about 9 months without issues (9 months ago I upgraded the whole system – so it didn’t really have problems then either… I was just randomly updating things).
Which one do you think I should measure the uptime on? The laptop is the one I physically put my hands on more often… but the server is more critical to my workflow.
My main Linux Desktop is actually a Lenovo x220 laptop running Fedora docked to a 30" monitor, a Kensington Expert Mouse and a custom Unicomp keyboard. It is on pretty much 24/7 so overnight backups and other jobs can run. I use suspend/resume when travelling because it works pretty well now. Uptime of usually up to a month is normal for me. Currently at only 3 days due to me switching kernels to fix an Intel GPU issue.
I also have a Linux server here which does things like Hg->GitHub mirroring for NGINX’s repositories. That has been up 35 days (power failure and I have no UPS). One of my headless Xeon workstations is also switched on right now for something I was doing over the last couple of weeks, 12 days uptime there.
I made myself a cheap desktop a few years ago because I couldn’t find any laptop in Taiwan with good Linux support. I try not to power it off and use the suspend function instead. It works like a charm: a press on the keyboard button and everything is back in action. Sometimes though, I have to reboot it, and apparently the last time was 15 days ago (a new Linux Kernel and Ubuntu asked me to reboot? I don’t remember)
On laptops, I tend to use the suspend function a lot, since it’s really faster to start at the beginning of the day. @bryanlunduke, what config do you have so that your laptop only takes 7 seconds to boot?
If it wasn’t because of a nasty bug that makes hibernating my main Linux system (aka my Arch-powered laptop) absolutely unreliable, I would just keep on hibernating it and not shutting it down daily… My server’s uptime is 12:29:07 up 23 days, 13:25, though
Interesting results so far. “Obviously” lots of daily shutter-downers and weekly rebooters (from a statistically significant sample of 19).
Of course after answering that poll, I’ve now jinxed my stats, and had to turn off power to the house (well that circuit, at least) to replace the light that decided to spontaneously stop working, because the halogen transformer melted.
My main Ubuntu desktop usually gets shut off every day, but it also has a super fast SSD so boot up times aren’t a concern anyway. It may stay on for a day or two max though when I forget to turn it off.
systemd has caused me to back away from Linux in favor of BSD. Actually systemd made me BSD-curious, but it was ZFS (among other things) that has kept me here.
Now having said that, I am running PC-BSD, which has the entire system running on ZFS. The nice thing about it is that it introduces the concept of boot environments. Thus, if the system updates, the first thing it does is creates a snapshot, and applies the updates against that snapshot. Thus, you don’t have an app fail because an underlying library got updated, and you don’t get the dreaded “Linux Daemons with Broken Links” finding in Nessus (a critical). And because the updates are in a snapshot, you don’t get any changes until you reboot onto the new boot environment.
The nice thing about this is that if something goes sideways, you can always roll back to the previous BE. I had this happen with a test box that was using the nvidia legacy driver, but somehow got upgraded to nvidia-current. I was able to roll back to the previous state to troubleshoot.
I boot/shutdown my laptop daily. It’s Ubuntu MATE on an SSD so blazingly fast to get going even though the hardware’s far from new.
My (Open Media Vault, so Debian) home server doesn’t get rebooted except when it kernel panics. Which is more often than it should, possibly because I’m a tight git and run everything other than the storage volumes off a USB drive caddy I happened to have lying around. This seems to be unpopular with the swap partition. Guess I should get around to fixing that
I’m exactly the same as @hillsy. SSD on my laptop so just switch off when it’s done, same with my Steam box. My server is on all the time and has a UPS to get it through powercuts, but we’re not comparing servers
Spoiler alert: nobody cares. Until ksplice is more commonly deployed and supported, if your uptime exceeds the date of the last security update to your kernel, then you’re potentially asking for trouble. I’m stopping short of saying “you’re an idiot”, because I wuv you guys and gals, but seriously.
In saying that, I shutdown an Ubuntu 6.06 server today with 998 days and 18 and a half hours of uptime on it. I feel like I’ve just managed to cremate an infected zombie so that they won’t bother anyone else ever again.
DISTRIB_DESCRIPTION="Ubuntu 6.06.1 LTS"
root@wuk1:~# uptime
17:33:16 up 998 days, 18:31, 2 users, load average: 0.00, 0.01, 0.05
root@wuk1:~# shutdown -h -f now
Broadcast message from root@wuk1 (pts/2) (Tue Apr 5 17:33:22 2016):
The system is going down for system halt NOW!
root@wuk1:~# Connection to wuk1 closed by remote host.````
While Ksplice has been around for many years, live kernel patching is now in mainline (with Red Hat working on kpatch and SUSE on kGraft). That should make this ability widely deployed in the near future.
