Server Maintenance

[dougjl]

I have kind of an odd question, something that seems should be obvious but really doesn't seem to be. It seems I spend most of my time fixing errors that occur on the server and never performing any kind of maintenance until I have to.

So My question is this, what procedures should be performed on an NT server as routine maintenance and how often? Should de-frag be run? What other utilities if any should be done on some kind of schedule? Of course a current backup is kept but that seems more like safety than maintenance. Any suggestons or ideas would be greeatly appriciated.

Thanks
Doug

 

[Zelandakh]

Depends what the server is doing. If its just a file and print server, there's not a great deal that needs doing.

You seem to be saying you have a lot of errors - are you running anything unusual?

I have perfmon running all the time on a graph showing network card in/out and proc usage. Set it to update every hour and check it every few days. Its a good indicator of the server health / spec.

Other than that my servers just run normally and tend to have around a one year up time. And they are clones running softRAID...

 

[DaveNamou]

Well Doug, I'm sorry to say, but I don't believe there is anything out there that will point out all the counter measures for errors and burps in the system that works for everyone. I feel that's it's something that you build on as time goes by that is unique to your own environment.

What I mean by that for example is this...
-My exchange server used to bomb out once every few months. It was the same error and the same fix every time. So, I wanted to track the event whenever it occurred and execute a script when the error occurred. This way, even though the problem occurred, it was automatically fixed within minutes by a script.

Now, there are common tasks that everyone should run such as
-Check the Event Viewer
-Defrag
-Backup
-ERD's
-Check on free disk space
-Disk Clean up
-Account Clean up (delete or remove unused user and computer accounts)

After a long period of time, you should have a nice list of things to do that is unique to your site. Your list should grow fast at first, but after a while you should only have to add things to it once in a great while.

Good Luck, hope that helps. Dave Namou, MCSE CCEA

 

[yizhar]

HI.

To be more specific about backup here are some todo's:


ERD (as Dave Namou wrote) - Run "RDISK /S" every month and keep the diskette. tip: if rdisk fails formatting the diskette, temporarly disable anti virus software.


Restore - you do backup to restore, right?
You should make a small and full test restore:
* small: restore specific folder from tape to server.
If you have Exchange - restore specific mailbox using brick level.
* Full - try to do "disaster recovery" on a test server,
or by scheduling a one-day of server downtime and restoring to a formatted hard disk on the server itself (keeping the originals disconnected in a safe place ofcourse).


It is very good to have a GHOST image of the server OS, ready for emergency.


You can also consider having a backup of important data on a workstation hard disk. This data can be accessed faster then data on tape in case of..


Bye
Yizhar Hurwitz

http://come.to/yizhar

http://teachers.sivan.co.il/yizhar