sanjay nadkarni wrote:
> Lori Alt wrote:
>> sanjay nadkarni wrote:
>>> Sorry for the delay in replying.
>>>
>>>
>>> Sarah Jelinek wrote:
>>>> Hi Sanjay,
>>>>
>>>> A few questions...
>>>>
>>>> Under Assumptions:
>>>>>  Since OpenSolaris supports only ZFS root file system, this is the 
>>>>> only root file system that is
>>>>>  supported. In addition for zones only ZFS zone roots are supported.
>>>>
>>>> For clarification, does this mean we only copy data from the root 
>>>> pool, or root pools if there are multiple(and allowed for multiple 
>>>> root pools)? The reason I ask is that the ZFS root filesystem, /, 
>>>> today does not have separate /var or /usr. I know that there is 
>>>> talk of allowing for a separate /var dataset, which means that this 
>>>> dataset could live in another zpool(not the root pool). So, is the 
>>>> plan to 'follow' any datasets that are part of the BE's? Even if it 
>>>> means spanning pools?
>>> I am not envisioning dataset  spanning pools.   S10 with ZFS root 
>>> support is also strongly recommending that zone roots be in the root 
>>> pool.   I this plan to stick with that.
>> We did recommend that at one time, but that was mainly because we 
>> could handle only so much complexity when we were introducing zfs root.
>>
>> Ed Pilatowicz and the zones team is now pushing us hard to relax this 
>> restriction and support zone roots on pools on files.  Yes, files. 
>> Either local, or exported from a file server (such as the 7000 
>> series) via NFS or CIFS.
> Interesting.  Do you have any info on the use case for this ?

yes, see:

http://bt2ws.central.sun.com/CrPrint?id=6915127

Look at the CR and the zoss project docs referenced in the CR.

lori

>
> -Sanjay
>
>> So I don't think that the "zone roots must be in the root pool" rule 
>> is going to stand.
>>
>> lori
>>>>
>>>>
>>>>> No user data payload. Flash provided the ability to include user 
>>>>> file systems. This was
>>>>> reasonable when disks were much smaller and data was closely tied 
>>>>> to the system. Today, most
>>>>> user data is either on a NAS or SAN device and they can be huge. 
>>>>> User data is often separately
>>>>> backed up. Hence it does not it does not make sense to include 
>>>>> user payload.
>>>>
>>>> I have been thinking about this, and I am assuming you mean that we 
>>>> won't copy anything outside of what is defined in the BE's, which 
>>>> right now is anything under rpool/ROOT/. So, /export/home is in the 
>>>> shared dataset space and would not be copied?
>>> That is correct.
>>>> I am asking because what does this mean for zones that have 
>>>> datasets as resources that are not in the root pool? Or a device as 
>>>> a resource. It sounds as if there is a possibility that some zones 
>>>> configurations will not be fully functional or bootable on the new 
>>>> system if we are only copying the content of the BE's.
>>>>
>>> Yup..but I don't quite see a way around this.  Perhaps the zones 
>>> team has some thoughts here.
>>>
>>>
>>>
>>> -Sanjay
>>>
>>>> Requirements:
>>>>
>>>>> Master images must support global and non-zones.
>>>>
>>>> We should probably outline in more detail what the specific 
>>>> requirements for non-global zones is.
>>>>
>>>> that's it for now.
>>>>
>>>> thanks,
>>>> sarah
>>>> ****
>>>> sanjay nadkarni (Laptop) wrote:
>>>>> Caimanics,
>>>>>    Please review the requirements documents for replication in 
>>>>> OpenSolaris.  This  provides the the  Flash type functionality but 
>>>>> includes support for zones.
>>>>>
>>>>> Thanks
>>>>>
>>>>> -Sanjay
>>>>>
>>>>>
>>>>> ------------------------------------------------------------------------ 
>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> 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
>>
>


Reply via email to