My mom used to say, “If you don’t have anything nice to say, don’t say anything at all.”  So, for my post about ACS yesterday, I feel the need to say that I flippin’ love ACS.

And now to the post…

A couple of weeks ago, I replaced the hard drive in one of our servers.  Well, we’ve had the joy this week of finding that the new hard drive has bad sectors.  Hooray.  So, tomorrow I’m taking the opportunity to switch over to a RAID 1 on that server with two new drives.  Still using 7200 drives though, not moving up to 10,000’s yet.  I’ll make sure to keep you updated.

Over the weekend, I did some maintenance on one of our servers. There were a few discoveries, and I have to admit that I am disappointed in myself for not discovering these things until now.

Over the past few weeks, I was hearing the tell-tale signs of a failing hard drive with the funny clicks and such. Running check disk confirmed there was a problem. I wasn’t too worried about the immediacy of the problem because I had been told by our previous contractor that the drives were mirrored and that the backup solution would work. Also, I had a replacement drive ready to be deployed.

I set up my schedule to be on site last Saturday (the 12th) to work on the server. It wasn’t long before I figured out that the drives weren’t mirrored at all, and in fact the supposed mirrored drive wasn’t even connected to power. I booted from that drive and discovered that it hadn’t been in use since 9/2006.

The backup system we are using is a $2,200 solution provided by our previous contractor that is a linear array of 4 500GB HDDs. (Ironically, also purchased in 9/2006) Each night, the server sends an image of it’s hard drives to the array located on the other side of a fire wall way on the other side of our building. Recovery is done by booting the server from a Linux restore disk. The disk failed to connect to the backup server using all methods I tried, but I verified that communication existed between the two. I found no helpful information about the specific error message on the 241 page manual, on Intel’s FAQs, or on Google. I will be contacting our previous contractor about this and also pursuing the Intel support department as to why we were unable to restore a backup.

In the end I used a tool to clone the existing drive to the new hard drive. It was done within 20 minutes. I then had to wrestle to get the drive letters back to the original assignments (and this took a lot of work for me). In the end, it was as simple as booting from a Windows 98 boot disk and running FDISK /MBR.

The moral of the story is to check and recheck your backup strategy to make sure you’re not left stranded.  If the disk had actually failed, I would have been in a world of hurt and the church would have been forced to fork out a lot of money to fix the problem.  So, I’m a dope, but I learned an invaluable lesson from this adventure.

I have never seen so many red X’s on an Event Viewer as I have recently on our main production server.  Looks like I’m going to have some fun times ahead.  I bet it’s a failing HDD.

Right now, our main backup solution isn’t able to create a reliable image of the drive.  We’re also having a number of services fail that don’t normally.  This server is utilizing a mirrored drive.

So, I hope it keeps on keeping on until the weekend so that I can take it offline with little impact.

So, I’ll divide readers with this post. Half will marvel at how dumb I am and the other half will be astonished by how simple the solution was.

So, I did a RAM upgrade on our sole server back on April 29th. It didn’t work out very well. Sure, nearly everything was great, but SQL Server 2005 wouldn’t connect to databases and had a list of cryptic error messages.

We have a guy on contract that’s been working with networks for a lot longer than I have been. He came in, put 4 characters in one file, and everything works swell. Go figure.

My initial error was Googling the cryptic SQL error messages, rather than searching by what I did… added RAM. So today I searched for Windows Server 2003 4GB RAM, and I came across Microsoft’s KB 283037. I’m curious now if the /PAE switch is a better one for us to use. Right now I’m not going to mess with it though - it’s working a TON better than it was before.

So, upon rebooting several of the machines, I realized that the group policy did indeed get applied on the successful restart.

Question to Microsoft: Why doesn’t the “gpupdate” command update ALL of the group policy settings? This may be a silly (and old) question to some, but I think that if you make a group policy update tool, it should update the group policy, all of it.

What I learned is that group policy relating to user settings apparantly updates after logon and group policy relating to computer settings apparantly updates before logon (and isn’t affected by gpupdate). I could be wrong, but that’s what looks like is playing out on these machines.

Hmmmm… a little rant for today, but I’m glad that it works. Now my network time is now synced via NTP (SNTP actually), and the users can’t change it.