It works only if the volume (either VxVM or ZFS) is exported as a single-slice disk (i.e. exported with the "slice" option). Otherwise if the volume is exported as a full disk then the accessible geometry is controlled by the disk label which is written on the disk.
There's a RFE opened for this: 6699271 (Dynamic virtual disk size management). alex. Stuart Davey wrote: > Hello all, > > Forgive me if I'm being stupid but I'm currently trying to grow a > Veritas Filesystem in a guest LDOM without any success. The Veritas > documentation suggests that it is possible: first grow the VxVM volume > in the control domain using "vxassist -g diskgroup growto volume length" > and then grow the file system in the guest domain using > "/usr/lib/fs/vxfs/fsadm -b num_blocks mount_point". The Vx docs state > that VxVM must be in the control domain and VxFS in the guest. > > The issue I seem to be having is that when I export the volume to the > guest LDOM, it sees it as a normal disk (i.e. 7 slices). Consequently I > go into format and partition it, label it and then apply a VxFS to it > before mounting. After growing the volume in the control domain the > guest domain still sees the original size volume (checked with format) > and so it can't grow the file system. Have I missed something obvious? > > Before starting out, I was hoping that I would simply be able to put > VxFS on a Vx volume in the control domain and then export it to the > guest domain - similar to what one would do with ZFS. My experience and > the Vx docs suggest that this is not possible. > > Has anyone out there been successful in growing a Vx file system in a > guest LDOM. If you have, and you can give me the "recipe" then I would > be very grateful. > > Thanks, > > Stuart > _______________________________________________ > ldoms-discuss mailing list > ldoms-discuss at opensolaris.org > http://mail.opensolaris.org/mailman/listinfo/ldoms-discuss
