> workaround_2 (as reported in the bug):
> - rebooting into single user, entering "maintenance
>   mode" as "root" and image-updating from there. 
> ------------------------------------------------------

this workaround is BAD as Evan just discovered:


While this does allow for a successful image-update it does not result
in zones that are usable. In the new global zone BE you will see the
zones and they will report that they are installed however attempting
to boot them will fail:
# zoneadm -z z2 boot
zone 'z2': ERROR: no active dataset.
zone 'z2':
zoneadm: zone 'z2': call to zoneadmd failed

This is because the image-update done in single user mode will not
create the zone datasets for the non-global zones in the new global
zone BE.

If you wish to maintain your zones into the new global zone BE then
workaround_3 is really the only way to do it. Yes this makes it so
your zones are no longer available in the old BE but they are available
within the new global zone BE.

- detach all of the zones on the system before the image-update.

This is also true for workaround_2. In both cases no new dataset or
BE gets created for the installed zone. However with workaround_3
it is at least possible to do a zoneadm attach -u to make the zone
accessible within the updated BE. Whereas with workaround_2 there
are additional steps needed to make the zones accessible in the
updated BE and the end result can be the same, the zones are not
accessible in the old global zone BE.
<snip end>
This message posted from opensolaris.org
zones-discuss mailing list

Reply via email to