James, Here's a story you'll like. I was working on our OS X Server last night. I did a full backup with QRecall, then installed the 10.4.10 update, then made a number of configuration changes, and the last thing I did before I left was another QRecall backup. Since I'm not sure where OS X Server stores all hundreds (thousands?) of configuration files, I didn't want to try to guess what items to back up, so I always have QRecall capture the entire boot volume. The server doesn't have much on it yet, so it goes pretty quickly.

Today, I spent the entire day playing with the server. As you know, I'm new to OS X Server, and Apple's docs just aren't that good. Neither are any of the books I've read. Maybe they're OK if you're a server expert, but I don't speak that language yet. So sometimes, when you're trying to figure out how to do something on the server, you just have to experiment.

So I'm setting up shared folders, adding a few users and groups, configuring DHCP, DNS, Open Directory, etc. Every time I found something useful, especially a configuration change, I carefully documented it via an entry in my weblog I started on the server.

Sometime this afternoon, things started to go south. First, shared folders which used to work stopped working. Then the help viewer application started acting up. (Links did nothing when I clicked on them, which meant I couldn't get past the first page of help.) I rebooted the server -- didn't help. I knew the help problem could be indicative of a cache problem, so I employed a cache cleaner utility, and rebooted. Sure enough, it fixed the help problem. But now I had a bigger problem. Suddenly, I wasn't authorized to make ANY changes in Workgroup Manager. I could authenticate against the directory, and got no error messages, but everything was grayed out.

Well, if I can't make changes in Workgroup Manager, the server is no better than an oversized boat anchor. So I fired up QRecall and started restoring things that I thought might have gotten hosed. Got nowhere with that either, so obviously I wasn't restoring the right things.

After 30 minutes of going nowhere, I decided it was time to get serious. I saved out a copy of my weblog page which documented the useful work I had done today. Then I rebooted the server off my external bootable backup drive. I ran QRecall and did a complete restore of the server's boot drive from last night's last backup. It took less than 30 minutes, and restored completely with no errors. I rebooted back to the newly restored drive, and everything works again! Most importantly, Workgroup Manager is speaking to me again. Sure, I lost the work I did today, but it didn't take long to restore that with the documentation from my weblog. (And we haven't rolled out the server to the users yet, so I didn't have to worry about users losing any of their files.)

I still don't know what I did wrong to cause the problem, but it's safe to say that if I hadn't had a good backup, I'd have spent hours, or even days, trying to fix the problem, and even then I might not have succeeded.

-- Bruce