[caiman-discuss] Slim Install: First Draft of Target Instantiation Design Spec

[Moinak Ghosh]

Ethan Quach wrote:
>
>
> Moinak Ghosh wrote:
>> Dave Miner wrote:
>>> Moinak Ghosh wrote:
>>>  
>>>> [...]
>>>>   The BE management function should handle pool configuration
>>>>   on the installed system. The Transfer module is active while still
>>>>   booted off the LiveCD. In this case the pool has to be temporarily
>>>>   mounted onto an alternate root and files copied over. The Transfer
>>>>   module will use zpool import -R ... to achieve this. This does not
>>>>   change any pool configuration. For eg. the root pool will still
>>>>   mount onto / when booted of harddisk.
>>>>   This temporary mount for the purposes of copying bits is easiest
>>>>   to handle inside the Transfer module itself.
>>>>
>>>>     
>>> I see no need for you to duplicate functions that the BE management 
>>> module seemingly needs to have.  Ethan, any comment?
>>>   
>>
>>    Should the BE management be responsible for doing a
>>    temporary alternate-root mount for copying files during
>>    install ?
>
> I see this functionality as being the equivalent of what we do today
> for "lumount", except that BE management will provide such interfaces
> from a library for applications like the installer.  The target root to
> install on is really no different from an alternate root when you're
> running off the Live CD.
>
> BE management will be available on the Live environment to do such
> things as BE discovery, but not for October.
>
> Thinking beyond Slim a little, the orchestrator will likely end up
> calling into BE management to "create a BE" rather than directly calling
> TI.  The orchestrator then calls into BE management to "mount the BE",
> then calls into the Transfer module to gets the bits laid down.
> Given this model, I think the Transfer module should expect to be given
> a mounted alt root dir rather than to have to do the work to mount
> itself.
>
> So for Slim, the stop gap should probably be that the orchestrator
> does the mounting after calling TI.

   Okay I get it now. Thanks for the explanation.

Regards,
Moinak.

>
>
> -ethan
>
>
>>
>> Regards,
>> Moinak.
>>
>>> Dave
>>>
>>> _______________________________________________
>>> caiman-discuss mailing list
>>> caiman-discuss at opensolaris.org
>>> http://mail.opensolaris.org/mailman/listinfo/caiman-discuss
>>>   
>>
>> _______________________________________________
>> caiman-discuss mailing list
>> caiman-discuss at opensolaris.org
>> http://mail.opensolaris.org/mailman/listinfo/caiman-discuss