Dave Miner wrote:
> On 02/11/10 10:25 PM, Jack Schwartz wrote:
>> Hi Dave.
>>
>> I want to clarify a response I made last night...
>>
>> Jack Schwartz wrote:
>>>
>>>>
>>>> Do the changes in ai_sparc_image.xml and ai_x86_image.xml resolve the
>>>> issue in 14054?
>>> I think so. I need to update the webrev with that bug ID. I believe
>>> 14054 is part of 13109. 13109 calls out two issues: making
>>> /var/pkg/state RW (14054) and the need to increase the size of the
>>> bootroot.
>>>
>>> aside: I don't think 14054 is a dup of 13109, as 14054 is under our
>>> control but the size issue of 13109 is an issue for the IPS team.
>> To clarify, 13109, which deals with installing using pkg, calls out two
>> issues:
>> 1) that /var/pkg needs to be RW (as corrected by Shawn Walker).
>> 2) that the bootroot now needs to be much larger due to how pkg now
>> works.
>>
>> As creators of image manifests, we are responsible for (1).
>>
>> The pkg team is responsible for mitigating larger pkg memory
>> requirements to handle (2).
>>
>> Assuming that pkg refresh uses the same catalogs as pkg install, then
>> 14054 would complete our part of what is needed to make that work. The
>> bootroot size issue still needs to be addressed, however, before the
>> complete fix is in place.
>>
>> By only modifying (1) the pkg refresh command still fails with "/ file
>> system full" in the middle of its operation.
>>
>
> A discrepancy here is that 13109 discusses the live CD issues, but
> your code review here is strictly about AI. What is the situation
> with AI, then?
Both live CD and AI boot into a ramdisk-based root filesystem, so the
problems are the same.
I tried the pkg refresh command with a "regular-sized" (unpadded) AI
image which had /var/pkg set to RW.
Thanks,
Jack
>
> Dave
>
