Now, shut down the system. Before you re-IPL, you will need to execute the following commands from VM:
LINK * 155 152
This fools VM into thinking that your 155-disk (the new /boot) is really at address 152. At that point, you can simply IPL 152 as usual, and you will come up in your new system. To get back to the old system, you can simply put the minidisks back the way they were.
You can perform a similar trick under LPARs, but with the following caveats: Since you can’t reassign device addresses, you need to actually IPL from the volume containing the newer image. Since you can’t just pretend the new disk is the old disk, you can’t automatically mount /boot in /etc/fstab. In addition, since you don’t have VM carving up your volumes for you, you’ll have to waste a lot of space by dedicating an entire volume to a few megabytes of boot code.
Although this technique is effective, it’s also a pain to implement. Let’s look at some better ways to do this.
INTRODUCING LELAND LUCIUS
All of the neat tricks we’re going to use in this article are the product of Leland Lucius at Ecolab in Minneapolis, MN. Lucius’ Website is www.homerow. net and is well worth visiting. Here, we will specifically be using his VMPARMS and LOADPARM patches, as well as a tiny patch to convert passed parameters to uppercase, so that we can actually use the VM PARM patch from the CMS command line.
BUILDING AND INSTALLING A NEW KERNEL