Summer Projects

Currently working 50 to 60 hour weeks on our summer projects. We are reimaging and migrating 6,000 Macbooks to OS X 10.5.3 this summer. I have two mobile image servers set up, and configured them to work with the Casper Suite from JAMF Software. I have two Intel Xserves configured as mobile image servers 1 and 2. I set each of them behind a router running NAT so they can still go over the WAN and hit the JSS and grab auto run data, but at the same time keep them off of any managed switches. This also means we can run a pure gigabit connection, since not all the switches at all the buildings are gigabit, we still have quite a few 10/100 switches out there. I also didn't want to deal with any type of over head you may get from a managed switch. Some people will cry that port fast has to be enabled to make this work properly, but it doesn't. When the Mac netboots it gets an IP rather quickly from just a regular old consumer router. I have already imaged several hundred Macbooks using this method. I have ran into some problems and performance issues. Most of it was with an older G5 PPC Xserve, which I have gotten rid of. The Xeons work so much better and faster. I also have some problems with the windows side and have taken a few steps to automate failures in a more easy manner. I just want to come out right now and say, with every technology there is a failure rate, and since mass imaging Macs with Windows on the image isn't exactly 100% fully native supported, there are some issues. I was seeing a failure rate as random as 25% to 0% each run. We had some runs with no failures and some runs with near half failures. My guess is that the open source tools used to write out NTFS may glitch here or there when mass imaging. Especially when unicasting to 20 clients at a time. So, to help these failures get reimaged right away in a more efficient way I put a little shell script in the command path of my netboot image installer. This script will wipe all partitions, and rename the drive "Macintosh HD." This is important because auto run data looks for that volume and the scripts we use to automate the imaging process rely on no partition schema at all, so it can create a new drive slice. If you have failed Windows partition on there, it will fail because the drive slices will be different. So, this is the script I wrote:   #!/bin/sh #if imaging fails toss this script into your executable path #example would be /usr/sbin #make sure permissions are set correctly, that it is executable #this will erase the disk and rename the volume Macintosh HD /usr/sbin/diskutil eraseDisk "HFS+" "Macintosh HD" disk0 echo "earsed boot volume, renamed Macintosh HD"   Then you save the script as wipehd.command and make it executable by doing a chmod +x /path/to/script. Then mount your netboot image under /Library/Netboot/NetbootSP0/Casper.nbi/install.dmg. Once you mount it, use the go to folder option and type the path: /Volumes/Install/usr/sbin and drop your new script in there. Now, if there is a failure you can quit Casper when netbooted and it will drop you into terminal. In terminal now just type wipehd.command and it will automatically wipe the drive for you. Which makes it way easier than opening the disk utility manually and using the GUI to wipe the drive out. I will eventually post a full article with pictures, AFP throughput graphs, configurations, and documentation on how to create good images. Once I get the time to put that all together that is.