launchpad-dev team mailing list archive

[Edwin Grubbs <edwin.grubbs@xxxxxxxxxxxxx>]

On Tue, Oct 12, 2010 at 3:46 PM, Edwin Grubbs
<edwin.grubbs@xxxxxxxxxxxxx> wrote:
> On Mon, Oct 11, 2010 at 2:29 PM, Robert Collins
> <robert.collins@xxxxxxxxxxxxx> wrote:
>> On Tue, Oct 12, 2010 at 7:54 AM, Edwin Grubbs
>> <edwin.grubbs@xxxxxxxxxxxxx> wrote:
>>> Bug 427263 (Impossible to edit bug descriptions with Konqueror) is
>>> currently tagged qa-bad. Does this need to be added to
>>> https://dev.launchpad.net/CurrentRolloutBlockers ?
>>
>> https://devpad.canonical.com/~lpqateam/qa_reports/deployment-stable.html
>> and
>> https://devpad.canonical.com/~lpqateam/qa_reports/deployment-db-stable.html
>>
>> are both authoritative for deployments these days; we can't deploy
>> anything at the moment - we're stalled on QA.
>
>
> I think these two reports may have exposed a serious problem with
> handling QA with the qa-needstesting tag. When we had a bunch of bugs
> with qa-needstesting, it didn't surprise me that the above reports
> showed that deployment is blocked on untested bugs. However, we now
> only have one bug with qa-needstesting, and the reports for both
> stable and db-stable show revisions that have no qa tags at all. Since
> the reports only show the oldest revision that hasn't been QA'ed, I
> have no idea if this is just a couple of unusual bugs, or if there are
> a bunch that we have missed.
>
> -Edwin


Now that I have looked at it more closely. It looks like the qa-tagger
report is pulling some bogus data out of the revisions. For example,
revision 11684 of devel says that it fixes bug 1544, but it is
obviously not related to that bug. Since launchpad doesn't even a show
a link between that revision and bug, I assume that qa-tagger is
getting the fixes list from the revision metadata, which is added with
"bzr commit --fixes". There are also problems when qatagger checks a
bogus bug number listed in the pqm message since there is no way to
unlink that. lifeless and I have opened bugs 659618 and 659629 to
track the problems with qatagger.

The good news is that we have been able to get the reports to display
the correct status by removing incorrect bug links and by adding
qa-untestable to the bug listed in the pqm commit message. I'll remove
that tag after the release.

-Edwin



>>> We still have several bugs that need to be QA'ed. Since staging is
>>> currently unavailable due to problems updating the db (see Tom
>>> Haddon's email), I have split up the bugs that can be QA'ed now on
>>> EDGE versus those that will have to wait until staging is working.
>>>
>>> Bugs that can be QA'ed on EDGE
>>> ------------------------------------------------------------------------
>>>    645403 TooNewRecipeFormat raised editing a recipe
>>>        ON EDGE ~rockstar/launchpad/recipe-too-new
>>>             merged into ~launchpad-pqm/launchpad/devel at rev 11612
>>
>> This is one of them :)
>>
>>>    645702 oops in holdMessage storing large message
>>>        ON EDGE ~sinzui/launchpad/hold-message
>>>             merged into ~launchpad-pqm/launchpad/devel at rev 11685
>>>
>>>    652626 Windmill egg needs to be upgraded to fix test breakages
>>>        ON EDGE ~wallyworld/launchpad/windmill-1.3r1544
>>>             merged into ~launchpad-pqm/launchpad/devel at rev 11684
>>
>> This doesn't have any bearing on deployment, I'll mark it ok (because
>> the test suite which it affects is working).
>>
>>>    627701 Make it possible to use feature flags to override the
>>> global timeout for specific pages
>>>        ON EDGE ~matsubara/launchpad/fix-bug-627701-override-global-timeout
>>>
>>>        ON EDGE ~lifeless/launchpad/bug-627701
>>>             merged into ~launchpad-pqm/launchpad/devel at rev 11668
>>
>> This cannot be QA'ed on edge. (The way to QA it is to drop the timeout
>> way low, or way high on a page that times out) - both will have
>> negative impact on our users).
>>
>> -Rob
>>
>