To reboot or not to reboot, that is the question?

Bleek

Bleek

Bleek

Soldato
Joined
7 May 2004
Posts
5,503
Location
Naked and afraid
As per the topic title.

One of our engineers has a task to create a reboot schedule for EVERY Windows based server on the domain, of which there are 250+.

My argument is that Exchange and Domain Controllers need not be touched as they either work, or they don't, and in fact that rebooting DC's too often can actually cause more hassle than it's worth because of service restarts etc.

Do you agree with me, or do you think/believe/know that these servers should be regularly rebooted, say once a week?

Thanks in advance.
 
There should be no reason to reboot them other than to roll out updates. This should be often enough for windows servers, i see no reason to reboot them every week other than to satisfy the old opinion that they degrade over time.
 
I wouldn't. You're going to have to do it in a specific order as well (as ideally you would need the DC's up before the other servers, especially if there's service accounts that rely on authentication).

Personally I don't see the point.



M.
 
Unless the reboot is required as part of a patch or has come out of an RCA that a reboot is required every so many days to prevent a real problem then there is no requirement for down time.

Operating systems, applications and hardware are generally more stable than in the past so there is little requirement for regular down time.


An example of a reason for down time is, if the server has regular memory leaks which in-turn cause the server to crash, this happens every 6 months, so you bounce the server every 5 months to prevent an outage during business hours.

The only reason you would periodically take all your servers down at one time is possibly for something related to the fundamental requirements of the servers such as contractual reasons regarding power supply to the server farm or air con maintenance, etc.
 
Last edited:
Depends what they're running, I've had the displeasure of using an app in the past which slowly killed it's host server, scheduled weekly reboot kept in running flawlessly - didn't like having to do it but it was the lesser of the various evils...

In general no, but don't rule it out in very specific proven cases.
 
Servers are something that are NOT supposed to be rebooted - something bad happens, just restart the appropriate service.

Windows degradation is effectively a thing of the past now.
 
A public sector body I recently worked at does this. I think some people like to dress it up in the 'preventative maintenance' gown but it's really avoiding underlying issues which could and should be sorted.

User perception though...
 
Elric said:
No scripted reboots required. Who came up with the task in the first place?
Click to expand...

I believe the Microsoft Team Leader did...

I can appreciate reboots for 'certain' app servers that require it because of programming inefficiencies but like I said in my original post, and as I've relayed to they guys here, I disagree with DC reboots in particular.
 
Last edited:
No reboots done here, unless obviously there's a problem.

Scheduled maintenance is done on DC's/Exchange 2007 once a month, for patches, updates, etc on a Saturday (less disruption to users)
 
We have a few servers set to reboot once a week, but this is due to third party software that doesn't quite work how it should do.

I hate rebooting a server - it masks current problems whilst potentially creating new ones.
 
When I started where I work now, every server was rebooted every single night :eek:

That was fine until one of the older servers (our core business db no less) started coming up with PCI errors on bootup, obviously causing a few issues!

My boss was adamant he wanted to keep rebooting daily but when we adopted DoubleTake to an offsite location, we couldnt possibly be mirroring every single day, so I convinced him to reduce the number of reboots.

Now, critical updates are pushed down from WSUS on a friday evening and they reboot at night. Over the weekend, doubletake mirrors, a full backup is take, a full AV scan is done, as is a defrag. The only exception is the Citrix servers, Citrix recommend rebooting them on a schedule (there's even an option to set that up through the management console) so they are done every 3 days.

It still seems a bit of overkill but a lot better than it was!
 
hes the leader so hes responsible, if hes happy with rebooting them then its his choice/head

He must see a benefit in doing it not matter how 90's the idea is at least there wont be any problems with getting down time to apply patches etc.:D
 
It doesn't matter what his position is, if he's wrong it won't be done. Simple.

Thanks for the input guys, you've cemented my thoughts and feelings on this - I've had a lot of servers removed from the reboot schedule as a result.
 
interesting, i assumed he had overall responsibility for the servers as he setup a task for all of them, if rebooting them once a week isnt suitable for your environment then heck yeah its wrong. Its also wrong if he was allowed to do this in the first place is there any change control in place? or are engineers running wild in there?
 
Not submitting a Change Request is a disciplinary offence and quite rightly so!

He's responsible, but not overall and is superseded by several IT Managers above. The current reboot schedule is a WIP and therefore not implemented, but since this thread I've made recommendations (diplomatic way of putting it) to remove the DC's and Exchange from reboots.

My point above was, if someone is wrong, they're wrong and it matters not their title or status they need to swallow the error and get on with things. In my world admitting a mistake and working to correct it is more admirable and commands more respect than battling on regardless through arrogance. :)
 
You must log in or register to reply here.
Back
Top Bottom