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

I was just visiting a server at the datacenter and notied that it has been up(without a reboot) since last October (11 months). It is a VMHost running Windows Server 2003.

Even if something has been running fine that long, should I reboot it?
What make makes you decide it is time to reboot a server?
Even if a server has been running for a year, would it just add an unneccesary toll on the components? How long have you had a server running without a reboot(especially one running Windows) ?

I just wondering everyone else's opinion.

share|improve this question

8 Answers 8

up vote 2 down vote accepted

I'll occasionally reboot my servers as part of testing the disaster recovery plan -- if there's a power failure, I want to be sure that my server will still boot up and be functional autonomously (or, if not autonomous, that the "Bring the server up" instructions still work as written). This especially after I've installed a new service that's supposed to load on startup, or modified one of the existing startup scripts.

And, of course, whenever I have to open the box to install new hardware.

Other than that, unless you're showing obvious slowdowns or an increase in system errors or anything of the ilk, rebooting is pretty much unnecessary.

share|improve this answer

Personally I'm not a fan of rebooting servers (except my terminal servers) unless there's a need for it.

share|improve this answer
4  
When patches come out, there's a need for it. –  David Mackintosh Sep 3 '09 at 2:42
    
Why the terminal servers? What makes them rebootable? –  BLAKE Sep 3 '09 at 19:23
    
Because they get hammered during the day running user processes, chewing up RAM and CPU. We reboot them every night as a "refresh". –  joeqwerty Sep 3 '09 at 19:30

I try to do a service window once a month. This is usually close enough to "patch release day" that I'm not too likely to get compromised, but far enough away that if there are going to be problems with the pending patches, some other sucker admin team gets to find them.

When it comes to Windows servers, my patch plan always includes two or three reboots:

  • one to clear the memory and DLL leaks and other general state crap that accumulates on a running computer
  • one post-patch reboot to complete the patching process (if required)
  • one fully independent shutdown and restart to prove that the computer will, in fact, start from cold in the current configuration.

Unix systems are usually a bit friendlier in I can usually skip the first reboot when doing updates on them and they rarely "require" a reboot after the fact, but I always do the shutdown and restart from cold.

Sure, I'm paranoid, but there's been good reason for that.

share|improve this answer

A server that has an uptime of 11 months is a server that is probably 11 months out-of-date with current security hotfixes (and other patches, updates and service packs). Having been up so long, I would highly recommend rebooting the system prior to installing any patches, so that if the machine does not boot up, then at least you know that it was not one of the many patches that broke its ability to boot. I would also suggest applying the patches in smaller sets, so that if one of the patches does break the server, then at least you have a smaller set to wade through to figure out what the culprit was.

share|improve this answer

Back in the bad old days of NT 3.5 and Novell it was usually a wise idea to reboot regularly, but in this day and age it's not so important.

We had an internal web server (Windows 2003 & IIS) online for 600 days before we rebooted it. Of course, 600 days without a reboot also pretty much means 600 days without any patching. And that's not such a smart idea.

share|improve this answer

I worked for a guy that ran Windows NT 4.0 for over 2 years without a reboot. Personally, I let Windows Update accumulate 'enough' patches, and then I go ahead and install. Probably go about 2 months.

share|improve this answer

The various servers I am in charge of there are 3 that I don't reboot on a quasi-schedule DCs 1&2 and our Exchange server. Everything else can pretty much go up or down after 5pm but before 10pm.

I do it mostly because of windows' patches and cleaning out the crusty stuff. Our Citrix ICA server, has the most set schedule of reboots.

share|improve this answer

I had a network from a somewhat failed company live on as a zombie (no one really wanted to take down the website and stuff). Last I checked, the DC and SQL Server has been up for 3 years without issue. The webserver goes down one or twice a year for patching.

There is no reason any modern server should get 'crusty' and be reboot 'because it's time'. If you have a server like this, try instead to find out what program/settings is causing that issue. I recently had an issue where the firewall software was leaking handles somehow and started causing issues, all the sudden, a server that needed to be rebooted once a month now is well past a month without any signs of issue. =)

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.