Dave Bevans wrote:
I have a customer with the following zones questions....

Originally my system had no swap partition, only “/”, “/var”, “/opt” and the extra filesystems for zone (/zone/z1, /zone/z2, /zone/z3). None of the zone filesystems were modified. Since Solaris was complaining about swap, I resliced the disks and recreated the metadevices without a reboot. /usr was separated from /, /opt was added back into /, and a swap partition was added. Although the production DNS zone is still running, I receive the following error for login:

*bash-3.00# zlogin dns*

*[Connected to zone 'dns' pts/2]*

*zlogin: exec failure: No such file or directory*

*[Connection to zone 'dns' pts/2 closed]*

I think you've wandered off into territory that has not been tested. Someone else may know what will happen, but my guess is that rebooting the system will fix the problem.

The behavior you describe is not surprising. The container has (well, 'had' :-) ) a LOFS mount into the GZ's /usr directory. When you moved the GZ's /usr, the container was pointing at invalid data.

My question is will a reboot of the zone fix the issues with the zone? I am assuming that unlike the real Solaris instance (the global zone), that the DNS zone could not tolerate the global zone filesystems unmounting and remounting. Assistance is appreciated.

Also, could we determine if there is any way to duplicate a zone using UFS/SVM? I would like to test changes to a zone before making changes in the future… and possibly to recover from this.

Solaris 10 will have this ability in the next update, late this year. It's *really* fast compared to other methods. :-)

Jeff VICTOR              Sun Microsystems            jeff.victor @ sun.com
OS Ambassador            Sr. Technical Specialist
Solaris 10 Zones FAQ:    http://www.opensolaris.org/os/community/zones/faq
zones-discuss mailing list

Reply via email to