Linux Format logo

How often do you reboot your server(s)?

Linux has a reputation as a stable operating system that doesn't often need rebooting. So we were wondering, how often do you boot your systems, and why?

We're mainly interested in figures for server systems, not desktops, and we realise that some intelligent guesswork may be needed to fill in the answers. Just give a reasonable estimate. If we get a decent response we'll publish the results in Linux Format.

Survey is now closed - thanks to everyone who took part!


Your comments

RHSE vs SLES

We rebooted our servers - core system or storage, whatever - only when it hangs, not because of maintenance tasks, reconfigurations and so. It doesn't matter what OS is about, RHSE or SLES, same reasons stands for both systems.

Linux servers only

I submitted my Linux servers details but they are a lot different to my Windows 2003 Server Domain Controller, it needs a reboot every time an update takes place LOL

Linux home server and VPS's

I have a CentOS server here at home for my NAS, and the only time that thing has ever been rebooted by me (meaning when it wasn't due to a power outage) was when I physically moved. It's very very stable, and tends to just work. Of course, I tread lightly on this machine - it's a single purpose machine.

My VPS's (linode and slicehost both) are the same story - they only go down when the Xen hosts go down by the hosting company. Otherwise, my uptimes are typically months on end...

Probably once every 50 days

Probably once every 50 days or so since that is about how often dapper gets kernel upgrades.

about once every 10 mins!

i keep rebooting mine cos its just there cos i can have it there but when i can i will get it working properly

Reboot, NEVER

Only time I reboot my server is to deploy, I maintian 2 laptop servers as a USB CDMA to wireless access point server via firestarter. Only other time I had to reboot before 300 days is when 1 system overheated due to dust build up from being left alone for so long (lazy me ....)

Up since the site is up - over 600 days

We never reboot our production servers. One of our server is up for over 600 days now.

Usually on for kernel updates

However, the last time I rebooted it was because it a Joomla library had managed to cause the server to hang - I had to phone 1and1 and get them to physically reboot it. The time before that was human error - I typed "shutdown -r now" in the wrong terminal (oops).

I forgot about the two content filtering proxy servers I run - but that's basically because I rarely have cause to touch them. One is reset every night and left off for an hour (cron combined with timer switch) to give it chance to cool down, it is a very old machine and lives in a hot room. The other has probably been up for over a year.

Engineering Manager

I have used Windows 2000 Server, Windows 2003 Server and Suse Linux systems at our Company. The windows systems are not robust and fall over on a weekly basis. This is usually after an update but can be random, due to email server or fax dying on the spot or some other weird and wonderful windows DLL. The email server was so unreliable I had no choice but to use a third party email application, Kerio Mailserver, on a Suse box for corporate email that works 100% reliably. I have never needed to reboot the Suse box in 18 months!! The Windows 2000 server also went the way of the dodo in the last few months. Guess which is next ;-)

Developer

Using XP Pro with Xitami webserver running as a service, so
I guess it qualifies as a server of sorts.

Using it as a multimedia kiosk server and uptime so far is 197 days.

No. 4 on counter.li.org

I had an Internet facing machine running SuSe Linux with kernel 2.6.5-7 with uptime 1315.9 days. It was No.4 on http://counter.li.org/reports/uptimestats.php.
Unfortunately there was a huge power outage in the datacentre in July, so it was restarted :-(.
Simply there wasn't any reason to restart it....

Whenever it locks

I only reboot mine for kernel updates and hardlocks. My home systems usually average about 3-4 weeks, my coloed servers are usually 3-4 months.

System Engineer

Our servers uptime:

1:34pm up 8 days
1:34pm up 168 days
1:34pm up 178 days
1:34pm up 178 days
1:34pm up 99 days
1:34pm up 10 days
1:34pm up 52 days
1:34pm up 52 days
1:34pm up 24 days
1:34pm up 52 days
1:34pm up 295 days
1:34pm up 295 days
1:34pm up 295 days
1:34pm up 295 days
1:34pm up 295 days
1:34pm up 263 days
1:34pm up 263 days

An average of 161 days, the ones with 8 and 10 days uptime needed a firmware upgrade for the hardware and the one with 24 days a software driver upgrade that needed a reboot.

We don't reboot them if not needed, even for software upgrades but tonight we will do firmware upgrades on all 17 servers so the uptime count will be 0 again.

/Andreas

Xenon's server

up 3 days, 44 min
only because I had a kernel upgrade
usually up for a month or so.

ubuntu server
needs ksplice really

Kernel Updates Only

Centos servers x2. Rock solid, restart on Kernel updates or hardware failure.
Bloody marvellous, why waste 1000's when you get this performance for taking the time to learn a little.

oracle server

My company is running Oracle on redhat and and the test environment has been running for more thatn 250 days and the operating environment for more than 560 days.
Stable as a rock! Most of our windows servers are rebooted every month due to Windows update!

Often enough to drive the IT peeps mad....

OK, so I've had linux servers running for long times with no problems, but I've also had Windows servers running for a long time too.
If you upgraded your Linux distro just as much as Windows then you would find that they're very similar. Linux has the advantage that a lot of it's modules do no need the kernel to be restarted in order to apply updates, but that is the same with Windows. Most software just need the services restarting rather than the OS.
It's just a matter of knowing what you are doing.
I've had web-developers tell me we should host on linux/apache server as they don't get viruses......Didn't have the heart to tell them that they do, oh, and we already are!

Tech Coordinator

I've got one CentOS machine at work as an off-site backup for a library automation system that has been up 245 days. My home server running Fedora Core 4 (file & print for Windows clients) has been up for over 1100 days. They just work.

Programmer

Our stable systems are only rebooted once every 15-24 months. Usually due to an program leaking memory.

Never.....on purpose!

Never reboot my servers, the only time is when the electricity goes off and the ups powers them down or if there is hardware failure(oh and when we had the fire!). Just had to replace the a hard-drive in my backup server and they had been going continuous for 4 years. The longest running we had was our firewall (ipcop on an old 1GHz second-hand machine) which went on to run for nearly 300 days.

I've built and manage

I've built and manage several servers over the past few years - my own included - which has been running without a reboot for nearly 2 years (although I recall I had to restart it once after a power outage).

Of the others which have been running 3 years, 2.5 years, 18 months, 16 months and 2 months respectively the only one I've had to frequently reboot is the first one and that was due to a dust build up in the case which repeatedly caused it to overheat. The client took a vacuum to it and it's been fine ever since.

The most recent server is due a reboot (Ubunutu 9.04 Server x64) but purely due to some upgrades. I'll do it the next time I'm on site.

A client of mine was praising the fact that they've never had to reboot their Opensuse-based server compared to when they had Windows SBS 2003 - which needed a reboot at least once a month!

Um, what?

@Whenver anyones not looking... (not verified) - August 24, 2009 @ 7:18pm

"I've had web-developers tell me we should host on linux/apache server as they don't get viruses......Didn't have the heart to tell them that they do, oh, and we already are!"

Uh, no they don't... they might get Windows viruses uploaded to them but actually getting infected with a Linux virus? I'm not going to do the usual middleground "oh well they do exist and perhaps you could.." NO! Just no! If you have an infected Linux machine, then you need to replace your admins. Simple.

Loads of unpatched servers

Loads of unpatched servers here... Hacker must love you people.

admin

Windoze daily unfortunately.
CentOS - about a dozen servers, roughly 2-3 times/year and that's only so we can run a fsck :)

Yay for Linux!

every couple of months

although for a non internet facing CentOS 4 box here we're at 492 days and counting. It will be retired in a week. I have to confess I'm conflicted. I'm delighted that it has reached nearly 500 days, but I'm also pretty horrified at the number of kernel patches it has missed.

Ah well, ksplice for the next one.



Web hosting by UKFast