I was too quick to think LU solves all my problems now :) Another test system has local zones. These zones live in a ZFS pool, within a system hierarchy rooted at /zones (pool/zones).
When lucreating the new BE to upgrade this system, I saw that LU has copied all zoneroots to the new BE's UFS root. That is, it made a directory /zones in the new BE and tried to copy all local zones' directory trees there. It failed soon, because the global zone's BE root is too small for all local zones of the system. And I think it was a wrong thing for lucreate to attempt anyway: I thought LU should have snapshot'ed and perhaps cloned the zoneroots and luupgraded the local zones in-place (clones would allow for luactivate and fail over to old OS). Perhaps I was wrong to use such a simple command for lucreate in my case? # lucreate -n snv_103 -m /:/dev/md/dsk/d3:ufs -- This message posted from opensolaris.org
