This might seem like a great way to do it, but it depends what you are starting with. Do you have a NIM server for a starter?
What current resources do you have? If you have 4 local disks that you can boot from (and all your data elsewhere) it might be better to make an alternate boot pair. Assuming the local disks are hdisk0-3 and hdisk0 & 1 are on one controller and hdisk2 & 3 are on another, you would be best to first spread your risk. Make the rootvg from hdisk0 &
2 for instance so you are protected from a controller failure. If you need to move disk contents from one to another, use
migratepv, but be aware that this can take a long time. Additionally, you will need to recreate the boot image and correct the boot-list as stated by
ibmtech
If you can do this process and it leave you two disks free, then you can use the smit panels:-
- smit
- Software Installation and Maintenance
- Alternate Disk Installation
This should build an alternate boot pair that you can revert to should your upgrade/migration have undesirable effects.
After this, you should test that you can boot from the alternate pair. You will need to get into SMS during the power-on process for this.
Another simpler approach would be to create and test a
mksysb backup and restore to a separate machine. You would then be able to recover to your existing server if you need to. Of course you would need hardware to test this on, although good DR service suppliers will allow you testing time for this anyway. I'm presuming that you have DR provision already? Now is a good time to test it!
I'm hoping that I haven't scared you, and I'm open to being corrected if I'm wrong. I just don't want anyone to assume that an
unmirrorvg will give you a split off boot disk.
I hope that this gives you something to consider and that it helps in your planning.
Robin
Liverpool/Blackburn
UK