Good,

I'm glad this is already being worked on.
Thanks,

-paul


On 05/22/09 17:29, Ethan Quach wrote:
>
>
> Paul Neary wrote:
>> I had this problem below on build 111b which poses interesting  
>> questions for future compatibility of OSOL releases.
>>
>> o I install an AI server using 2009.06 CD
>> o I set up both SPARC and X86 clients using default manifests 
>> (pointing to pkg.opensolaris.org/release)
>> o The X86 client installed fine.
>> o The SPARC client croaked with
>> "...
>> ict_installboot Command /usr/bin/env -i PATH=/usr/bin 
>> /usr/sbin/installboot -F zfs /a/platform/sun4v/lib/fs/zfs/bootblk 
>> /dev/rdsk/c6t0d0s0 failed with 256
>> <OM May 19 19:43:57>
>> installboot failed
>>  ...
>> See full log file in attachments
>>
>> I guess that since 2008.11 didn't support AI on SPARC then I've 
>> inadvertently tried to do something that's not supported.  Also, this 
>> only happens right now because  pkg.opensolaris.org/release is 
>> pointing to 2008.11.   When we actually release 2009.06 this will 
>> point to 2009.06.
>>
>> The only way a customer will encounter this problem is if they have 
>> already cloned a 2008.11 repo and then use  2009.06 AI to install on 
>> sparc while pointing to their old repository.
>> Questions I have are,
>> 1. Should this fact be release noted for 2009.06?
>
> There is a note in the AI documentation stating that the
> build number or release being installed must match the build
> number or release of the AI iso image used to install it.
>
> Though it does not necessarily hint that the repository specified
> in the manifest can dictate the build being installed.   It only 
> refers to
> how the pkg list specification can be modified to dictate the build
> number.
>
> IMO the issue as stated doesn't need to be additionally release noted.
>
>> 2. Are 2009.06 AI and all subsequent releases going to be compatible 
>> with repositories from earlier releases (x86 and SPARC)? If not, 
>> deliberately checking for compatible/incompatible repositories and 
>> failing gracefully would be nicer than an obscure failure (as in this 
>> case).  Does this qualify as a possible RFE?
>
> This is bug 7837
>
>   http://defect.opensolaris.org/bz/show_bug.cgi?id=7837
>
> being actively worked on, project page here:
>
>   http://opensolaris.org/os/project/caiman/CVERS/
>
>
> thanks,
> -ethan
>
>


Reply via email to