[SunHELP] updating zones after installcluster to inactive BE

Richie listmail at triad.rr.com
Thu Jan 17 12:18:22 CST 2013

On 1/17/2013 11:02 AM, Richie wrote:
> On 1/17/2013 10:00 AM, Richie wrote:
>> On a test box I created a new BE and ran installcluster -B 
>> s10u10_CPU2013-01 --s10patchset
>> The new BE came up fine, but I forgot about implications of having a 
>> non-global zone.  I currently see a process "zoneadmd -z myzone -R 
>> /.alt.orig" which implies the old/original BE is mounted. I'm aware 
>> the old BE get mounted on new BE bootup so that lu can do some sort 
>> of a sync.  This has been out there for a while and i'm not sure if 
>> it is hung or waiting for me to do something.
>> Is there something I need to do so the old BE gets unmounted and to 
>> ensure the updates get applied to the zone?
>> I took this approach (instead of a zone detach and later reattach 
>> with -u option) in order to minimize downtime.
>> TIA
> Just an update.   The box is an old SF4800 running Solaris 10 update 
> 10 and zfs root.
> The verbose cluster log indicates that the zone also got patched so 
> that portion is not a concern.
> I chose to do a reboot and for whatever reason the old BE still gets 
> mounted.
> # ps -aef |grep .alt
>     root  2610  1156   0 10:55:16 pts/1       0:00 grep .alt
>     root  1269     1   0 10:51:52 ?           0:00 zoneadmd -z myzone 
> -R /.alt.orig
> # mount |grep .alt
> /.alt.orig on rpool/ROOT/s10s_u10wos_17b 
> read/write/setuid/devices/rstchown/nonbmand/exec/xattr/atime/dev=4010009 
> on Thu Jan 17 10:51:50 2013
> /.alt.orig/var/run on swap 
> read/write/setuid/devices/rstchown/xattr/dev=5e8000c on Thu Jan 17 
> 10:51:50 2013
> The zone is autoboot = false, but I have no problem starting/stopping 
> it on the new BE.

pardon me for continuing to talk to myself, but I found the core problem 
in /etc/lu/sync.log

ERROR: unable to mount zones. Error message follows.
ERROR: Unable to mount zone <myzone> in </.alt.orig>.
zoneadm: zone 'myzone': "/usr/lib/fs/lofs/mount 
/.alt.orig/zones/myzone-orig/lu/a/opt/sfw/etc" failed with exit code 111
zoneadm: zone 'myzone': call to zoneadmd failed

Looking at the old BE filesystem the zone dir is indeed renamed.
# ls -l /.alt.orig/zones
total 3
drwx------   5 root     root           6 Jan 17 08:57 myzone-orig

The above mount would never work, but on second glance the parameters 
are actually backwards.  I think the command should have read:
"/usr/lib/fs/lofs/mount /.alt.orig/zones/myzone-orig/lu/a/opt/sfw/etc 

More information about the SunHELP mailing list