Also I had to clean up some mess due to the fact that the actual boot disks are 
not the ones with the ZFS partitions, but after some semi-random actions snv_93 
is up and running on this server! :)

Now I'm trying to upgrade the zones, completely hosted on a zfs pool. I made 
another BE as a clone of the upgraded one, and put the zone descriptors back 
into /etc/zones.

Each zone has a zfs filesystem in a pool/zones/*/* tree (mounted as 
/zones/*/*), and some have delegated datasets, and these cause problems. Seems 
they are all parsed as one big string:

# lumount -n snv_93.tmp
ERROR: unable to mount zones:
cannot open 'pool/zones/cos/cosldap02
pool/zones/cos/cosldap02/localzfs
pool/zones/cos/cosldap02/localzfs/backups
pool/zones/cos/cosldap02/localzfs/backups/ldap
pool/zones/cos/cosldap02/localzfs/backups/ldap/dsins1': invalid dataset name
chmod: WARNING: can't access /.alt.snv_93.tmp/zones/cos/cosldap02
zoneadm: /.alt.snv_93.tmp/zones/cos/cosldap02: No such file or directory
could not verify zonepath /.alt.snv_93.tmp/zones/cos/cosldap02 because of the 
above errors.
zoneadm: zone cosldap02 failed to verify
ERROR: unable to mount zone <cosldap02> in </.alt.snv_93.tmp>
ERROR: unmounting partially mounted boot environment file systems
ERROR: cannot unmount <rootpool/rootfs_snv93.tmp>
ERROR: cannot mount boot environment by name <snv_93.tmp>


What's worse, such attempts leave around "/.alt.snv_93.tmp/var/run". It it 
mounted by kernel, not accessible to users or admins, can't be umounted (busy) 
and only gets fixed by a reboot. Not good.
 
 
This message posted from opensolaris.org
_______________________________________________
sysadmin-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/sysadmin-discuss

Reply via email to