I've read and read and read the docs, and I just don't get it, so I'm here to ask for clarification.
I have a T2000/Solaris10u8. It has three drives. Drive 0 and Drive 1 are mirrored with metaxxxxx. Drive 2 has some untouchable data. It hasn't been treated nicely in a very long time and now it's on my plate.
It needs to be patched. So my approach was to install a fourth drive (Drive 3) and do a lucreate to copy the system to a backup environment on Drive 3. I then patched the backup and luactivate(d) it and booted from it. Everything was fine. Since it's only one drive, it exists as c1t3d0s0.
So after successful testing, I want to return control to Drive 0 so I can take Drive 3 out and use it on the next machine to update. Lumake didn't like me trying to use either md0 or c1t0d0s0. In the end I got rid of the original BE that was on Drive 0 (ludelete) and did a lucreate from Drive 3 to Drive 0 (using its /dev/md/dsk/d0 address). When I activated Drive 0 and rebooted, it came up just fine. The mirror was in place and metastat reported all was good.
But that all seemed kind of a clunky way to do it. Anyone have a better, more educated and experienced approach?
You created this thread a week ago and no response so far. so here goes.
You mention a T2000 and a need to upgrade the next machine. You have a number of these?? Generally you can only insert a mirror copy in another machine if it's exactly the same hardware (or you are into the trickery necessary to make it work).
The T2000 has a built-in hardware raid controller so you can configure a RAID1 mirror of a drive. Unfortunately, to mirror the boot drive you need to create the mirror before you install the O/S. If you do that then you could pull out one leg of the mirror (ideally when the system is quiescent). The system will see this as a drive failure and you can insert a new replacement drive and it will rebuild. This way you can "breed" copies.
(I have found that sometimes you need to rewrite the bootblock after you insert it into another machine to make it boot.)
Unfortunately, I guess, I did not go the route of hardware raid since it was already RAID configured through the meta tools. I was very fortunate to have a spare machine that is a hardware twin to the one that needs to be update so I've been able to create the same environment for testing methods of cloning, etc.
I'll be doing a test clone of a second T2000 today and so far it has gone identically to the first one.
Okay, I see where you are going with this and why.
Just one tip though.........
You mention T2000, Solaris 10u8 and upgrading/patching. I've had lots of trouble with this due to OBP/Hypervisor/firmware being too old. I've had the application of patchsets crash repeatedly on T2000 but Oracle don't seem to flag it as an issue.
Search this forum for T2000 patching failure, or the like, and/or look through my posts and read the threads that I've contributed to on this subject.