Take the 2-minute tour ×
Server Fault is a question and answer site for professional system and network administrators. It's 100% free, no registration required.

We have a Windows server 2008 R2 being used as a mail server for less than 1000 users. I want to ask how frequently should the server reboot once ? If no reboot action is taken, how long normally will the server be standing ?

share|improve this question

6 Answers 6

up vote 2 down vote accepted

I'm the opposite of Joeqwerty. I believe that server reboots should be part of a maintenance plan. The longer that a system stays up without a reboot, the more changes could possibly be occurring when you finally do reboot. There's also the possibility that changes that required a reboot could be forgotten and then when the reboot does happen months later, unexpected changes occur.

Not to pimp my blog, but I explored it in depth in a post called "Epic Uptime – Bragging Rights or Epic Fail?" in which a lot of people who are much smarter than me weighed in and swung the general consensus to be that rebooting == good.

IMO, I would reboot the server about once per month, which would be normal anyway since patches are released once per month. I reboot after patches are applied regardless of if they say they require a reboot... especially if they're .NET updates. I apply each .NET update one at a time with a reboot in between and thus have never had a .NET update installation error on any production machine. Googling the KB numbers of any .NET update is replete with complaints about them refusing to install or constantly coming back up in the update list.

Just my proverbial two cents.

share|improve this answer
    
I know that this question is one that generally incites some pretty heated debates, but I'd like to see some actual data that supports the notion that reboots are an effective maintenance task and help to improve overall uptime, stability, etc. It seems that most of the arguments for rebooting are based on conjecture, hyperbole, and personal opinion and preference. Can you point me to some supporting data? –  joeqwerty Oct 27 '10 at 13:21
    
@joeqwerty I don't think there's been any empirical studies to show one way or the other. Right now it's just lore, anecdotes and theoretical arguments that support one way or the other. Even in my comment above I used an argument from authority; "A lot of people who are smarter than me weighed in...". Yeah, I suck. Perhaps we should work on making the first scientific study into the topic? =) –  Wesley Oct 27 '10 at 15:21
    
Good idea. I'll compile the stats and data from the "I don't reboot my servers" camp and you do the same from the "I do reboot my servers" camp. We'll meet back here in a week. ;) –  joeqwerty Oct 27 '10 at 15:45

I don't reboot servers unless there's a reason to do so. I'm not of the opinion that server reboots should be a part of any regularly occurring maintenance plan. If there's a problem, update, or install that requires a reboot then I do it, otherwise I don't.

share|improve this answer

Unless a service is badly behaved for some reason, the monthly Microsoft update-reboots is as much as my Windows servers need.

share|improve this answer
    
We organized our SLA maintenance windows around this... the Saturday morning following every 2nd Tuesday is permissible down time to apply server updates and reboot. –  Joel Coel Jan 28 '13 at 17:55

A server should only be rebooted when changes to it require a reboot. For example a patch or hardware change that requires a reboot.

If you find that you have to reboot more frequently than that then one of the applications you have installed is buggy or the hardware is faulty in some way, and the answer to that is to fix the bug/fault, not to set up a scheduled reboot task (though I appreciate you might need to do so while waiting for the bug/fault to be fixed).

share|improve this answer
    
Vote++. I'm struggling to get my head around the concept of rebooting a server as a routine measure. We spend a lot of time making sure the service we provide is fast and reliable. We achieve this by using reliable hardware, well managed O/S deployments and strict change control. I'd be interested to hear our customers' response if we introduced a regular reboot (something I last saw with weekly mainframe IPLs). –  Simon Catlin Oct 27 '10 at 21:19

Servers reboot when they need to be patched. Point. Any other reboot is either crap hardware (or better stupid drivers, to e replaced) or a broken server (as in defective hardware). Or really crappy software + an idiot sysadmin which restarts the server intead of the software only.

A full server (ie.e. no server core) will reboot something as monthly to every other month. Simply becasue patches demand it.

A server core install depends - it requires a lot less reboots and I think my Hyper-V only server once survived 6 months without requiring a reboot. every 2-4 months is the norm.

A server freezing or requiring regular reboots to work normally is somthing I normally handle via RMA lines (fix it or take the shit back) with the supplier. Normally a motherboard / RAM change fixes all stability issues.

share|improve this answer

It depends on your SLA's, rebooting should not be something you usually do, it must be done as part of a Maintenace plan, Emergency or security related procedure.

Here at my company we have a server restart manual that describes the order which we restart the servers, however we don't have a procedure that stablish the perodicity of server restarts.

share|improve this answer

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.