2 server backup

[rjmccorkle]

We now have 2 identical hardware servers. The orginal server, B001 (context UNIV), has all of our data. We would like to use the second server, B002, as an extreme emergency use only server; it would be unplugged from the electrical supply and network except for a brief weekly connection to update the NDS. So far, this appears possible. We installed NW5.1 SP6, using the existing tree on B001 while physically attached to the network. After DOWNing B002 then physically detaching it from the main network and creating a mini test network of the server B002 & 1 workstation, our original tree structure appears on B002. If we do NWADMIN changes such as add a User on B001 then re-attach B002 to the network, the tree structure seems to get updated on B002.
We created duplicate Volume and directory structures on B002 so hopefully we can use PKZip to backup data from B001 then restore it to B002. We have not purchased a second server license yet. In the event of server 1 going out completely, we plan to use the license disks that we used on server 1 to finish the license install on server 2.
2 Questions:

1. The orginal server, B001, constantly (about every 25 minutes) looks for the second server now, generating error messages. Is there a way to tell server B001 to stop looking for B002, but maintain the NDS synchronizaton when we attach B002?

2. Can B002 (the backup server) be renamed to B001 in the event of an emergency in a way that all of the file/directory rights & the printer queues will work from the original B001? (I'm guessing an alternative is to not rename B002, but recreate the print queues on B002 by hand)

Briefly, can we use the second server in place of a "tape backup"?

In the experiments I have done so far after setting up the 1 workstation network with B002 as the server, I can login to B002 with users/passwords established on B001. I tried just changing the server name in AUTOEXEC.NCF to B001. The drive mappings worked but every 5 seconds the server console says "Unable to communicate with server .B001.UNIV" . And the print queues don't work (I thought there would be some structures that my simple method would not copy).
Thanks for any help.
Bob

 

[Provogeek]

Bob,

Question 1: With out removing the second server from DS, there is no way to stop the first server from looking for it. You can tell the little bell to shut up by using the SET SOUND ALERTS = OFF command (could be wrong on syntax, don't use it very often).

Qestion 2: You can rename it, but it will screw up your DS in the long run since you have a server of the same name in the tree. Renaming a 5.x server doesn't have much impact on it (changing the IP does). Be sure you do not change the internal ID at the same time. Change the name, bounce, then change the internal ID (also called the internal IPX). Remember, print queue technology is legacy and out dated. It is bindery and not very well directry enabled. The queue created directries on the server and this is were clients place their print jobs. Your second server does not have these directories, that is why your print tests fail. Look into NDPS, it can do what you want. You can backup NDPS info to the tree so the NDPS Manager can be moved from server to server with out affecting the clients. Note: You don't need a broker to run the Manager, so it would work for the emergency, but you will need a broker to distribute drivers.

Using this method in place of a backup is doable, but not best practice. You are left with no history, and no ability to restore that file a user screwed up a month ago and just noticed it was messed up. It is actually better to leave a system up and running than it is to power on and off all the time. It's kinda like starting your car puts the worse wear on your engine, same goes for motors in hard drives.

You could look into this product to acomplish the same thing

http://developer.novell.com/yes/62963.htm

Not sure of avilability though, I don't see more info on the publishers web site, you may want to call them and ask.

=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+
Brent Schmidt Certified nut case
Senior Network Engineer

http://www.kiscc.com

 

[rjmccorkle]

Brent,
Thanks for the help. Over the last month I have talked with a couple of other Netware-experienced people and they echoed your advice, keep the backup server running. So I shall probably do that and make a few batch files for copying the few files we keep on the file server over to the backup server. My one and only experiment with NDPS worked after the initial setup but did not work the next day. The biggest problem I face is finding a grounded electrical outlet in one of the other buildings for the backup server. Thanks again for the help.
Bob

 

[terry712]

for ndps as long as you have a copy of the res dbase normally sys/ndir/res or something like that they you can load the agents on any server

i take it upgrading to 6 and installing the freebie ncs 1.6 isnt an option