Mick Atkins wrote, On 01/10/2009 11:50: > I have created a ufs dump archive using the latest ldmp2v of a sunfire v210 > running sol 10 11/06 > My Control ldom is sol 10 5/08 with latest recommended patches running on > t5120 v1.2 I can create guest ldoms with no issues. > I set up an nfs share and ran a ldmp2v collect which was successful. I > created a zpool and zfs dataset ldompool/edev4 > When I try ldmp2v prepare -v -d /destination edev4 it fails when the control > domain wants to reboot for a change in IO delayed reconfiguration.
Your control domain needs to be running the equivalent of Solaris 10 10/08 (S10U6) or greater (or patched to a level equivalent to 137137-09 or greater) This is because the P2V tool depends on Virtual I/O DR (which was introduced in 1.1/S10U6) and the delayed reconfiguration indicates that VIO DR isn't working. You mentioned that the latest patches are installed - can you check to see if 137137-09 (or a patch that obsoletes it) is installed. - Liam > If I let the control domain reboot then the ldmp2v prepare fails saying disks > already exist. It creates the disk0 file in the zfs dataset > My /etc/ldmp2v.conf is : > VDS="primary-vds0" > VSW="primary-vsw0" > VCC="primary-vcc0" > BACKEND_TYPE="file" > BACKEND_SPARSE="no" > BACKEND_PREFIX="/ldompool" > BOOT_TIMEOUT=60 > > I have tried changing my conf file > VDS="primary-vds0" > VSW="primary-vsw0" > VCC="primary-vcc0" > BACKEND_TYPE="zvol" > BACKEND_SPARSE="yes" > BACKEND_PREFIX="ldompool" > BOOT_TIMEOUT=60 > Then when I run I get : > Creating vdisks ... > Creating vdisk edev4-disk0 ... > Creating volume edev4-vol0 at primary-vds0 (70494 MB)... > invalid option 'p' > usage: > create [[-o property=value] ... ] <filesystem> > create [-s] [-b blocksize] [[-o property=value] ...] > -V <size> <volume> > > For the property list, run: zfs set|get > Cannot create ZFS volume ldompool/edev4/disk0. > Cannot create edev4-vol0 at primary-vds0. > > Can anyone point me in the direction of detailed ldmp2v documentation or tell > what I am doing wrong > Thanks > Mick
