Hi, William,
Would you please let me know the usage of the command line options,
is it adding "-i/I' to grub menu? or
/usr/bin/auto-install -i/I?
And which build can I get this fix? The coming build 110? I see the bug
status is fixinsource.
Great Thanks
Jeffrey
William Schumann ??:
> Jeffrey,
> Jeffrey Huang wrote:
>> ...
>>> Jeffrey,
>>> The fix for 5653, now in source, includes command line options to
>>> facilitate Automated Installer testing.
>>> -i will cause AI to exit before Target Instantiation
>>>
>> "Before TI", it is useful for us to test cases which verify various
>> criteria, since I only need to get the manifest
>> w/o TI and IPS.
>> Now what I am doing is given a fake target device to make the AI fails.
>> With "-i", I do not need to set this fake target device any more.
>>
>>
>>> -I after
>>>
>> So "After TI", does it means already physically changed the hard disk,
>> so the actual partition and slice
>> are created on disk, right?
>>
> YES
>> If YES, then it will be very useful, it could save us a lot of test
>> execution time, because now my test case
>> has to wait the whole installation process end to verify disk actions.
>>
>> And further more, I have a question:
>> Do "TI" create zpool and zfs? Do we have an option to disable creating
>> zfs dump and zfs swap dataset on rpool?
>>
> Not yet.
>> Because when I am trying to execute the test case again and again, I
>> have to export/destroy zpool from last AI,
>> but I have to remove dump and swap device from zpool by command:
>> # swap -d /dev/zvol/dsk/rpool/swap
>> # dumpadm -d swap
>>
>> but this 2 commands quite often cause whole ZFS hang (I guest zfs has
>> serious problems),
>> then I have to reboot the machine.
>>
> I am also forced to reboot sometimes when doing this in testing, and it
> would be a nice feature and not hard to implement.
>
> If you file a bug for this, we should be able to get it in, with 2009.1H
> being the highest priority at the moment.
> William
>>
>>> An exit status of zero indicates success, otherwise failure.
>>>
>>> Please let me know if this is useful to you.
>>> William
>>>
>>> jeffrey huang wrote:
>>>
>>>> Hi, Jan,
>>>>
>>>> So, that being the case, we will stop new development and putback of
>>>> liborchestrator test suite,
>>>> then put more efforts on AI test cases development to cover those
>>>> functions.
>>>>
>>>> Thanks
>>>>
>>>> Jeffrey
>>>>
>>>> jan damborsky ??:
>>>>
>>>>> Hi Jeffrey,
>>>>>
>>>>>
>>>>> On 02/13/09 08:48, jeffrey huang wrote:
>>>>>
>>>>>
>>>>>> Hi, William and Jan,
>>>>>>
>>>>>> I'd like to suggest that can we give up the liborchestrator test
>>>>>> suite? I am based on following reason:
>>>>>>
>>>>>> liborchestrator test suite based on 2 test driver:
>>>>>> - disktest
>>>>>> - kbdtest
>>>>>>
>>>>>> 1. disktest is unnecessary when AI manifest processing was
>>>>>> implemented, some of the features could be covered by AI testing and
>>>>>> TD testing
>>>>>>
>>>>>>
>>>>> I tend to agree. We have recently discussed this with William
>>>>> and he pointed this out as well - the same level of test coverage
>>>>> can be accomplished by combination of TD and AI - William
>>>>> suggested to enhance orchestrator for testing purposes
>>>>> by specifying set of checkpoints in which we could terminate
>>>>> the process and inspect test results.
>>>>>
>>>>>
>>>>>
>>>>>> 2. kbdtest is plan to be replaced with G11n test driver which would be
>>>>>> delivered with G11n module proposed by globalization team.
>>>>>>
>>>>>>
>>>>> kbdtest is definitely out of date and thus useless.
>>>>> As you pointed out, the plan was to replace it with
>>>>> the one delivered with G11n module. However,
>>>>> no ETA is available at the time being, so the plans
>>>>> might need to be accommodated - stay tuned :-).
>>>>>
>>>>> Thank you,
>>>>> Jan
>>>>>
>>>>>
>>>>>
>>>>>> 3. And the current test cases we could implemented and features we
>>>>>> could covered are quite limited.
>>>>>>
>>>>>> I came out above conclusion by tracking following bugs and email
>>>>>> thread for "code review request 5656":
>>>>>>
>>>>>> bug 6193 Needs to integrate test binaries to package when build
>>>>>> slim_source
>>>>>> bug 5644 TD and TI test drivers should be built as part of full build
>>>>>> of source gate
>>>>>> Bug 3112 liborchestrator test driver should be enhanced to address QA
>>>>>> requirements
>>>>>>
>>>>>> --
>>>>>> Great Thanks & Best Regards
>>>>>>
>>>>>> ---
>>>>>> Jeffrey Huang SWAN: x51683, Ext: 0086-10-62673683
>>>>>> Solaris QE Group Email: jeffrey.huang at sun.com
>>>>>> China Engineering & Research Institute, Sun Microsystems
>>>>>> 7/F Chuang Xin Plaza, Tsinghua Science Park, Beijing China
>>>>>>
>>>>>>
>>>>>>
>>>>>
>>>>>
>>>> --
>>>> Great Thanks & Best Regards
>>>>
>>>> ---
>>>> Jeffrey Huang SWAN: x51683, Ext: 0086-10-62673683
>>>> Solaris QE Group Email: jeffrey.huang at sun.com
>>>> China Engineering & Research Institute, Sun Microsystems
>>>> 7/F Chuang Xin Plaza, Tsinghua Science Park, Beijing China
>>>>
>>>>
>>
--
Great Thanks & Best Regards
---
Jeffrey Huang SWAN: x51683, Ext: 0086-10-62673683
Solaris QE Group Email: jeffrey.huang at sun.com
China Engineering & Research Institute, Sun Microsystems
7/F Chuang Xin Plaza, Tsinghua Science Park, Beijing China
