Thanks guys for fixing this.  This is actually the third time that
this has occurred for me.  The first was a couple months ago after the
big unexpected downtime, and then these two additional entities.
Unfortunately, this latest record relies upon the key name, so I
couldn't create a record of duplicate values. With my inability to
delete the record, I wouldn't have been able to free up the key name
for use in the new record. Also, this record has several dependencies,
which I could trace down and change, but I would prefer not to if at
all possible.

On Jul 14, 10:57 pm, Alfred Fuller <[email protected]>
wrote:
> Just to clarify this is an incredibly rare occurrence (lucky you :-)) and we
> will fix it permanently (it just takes a little man power). Changing the key
> name would fix it as well, but we don't expect you to have to do this.
>
> On Wed, Jul 14, 2010 at 7:44 PM, Ikai L (Google) <[email protected]> wrote:
>
>
>
> > Hi Daniel,
>
> > The sort order should be working correctly again. The issue was a broken
> > index value; we have a short term fix in place, but I wouldn't rule out this
> > appearing again. Is it possible for you to create a new entity with the same
> > values with a different key? That's one way to work around this issue in the
> > future.
>
> > On Wed, Jul 14, 2010 at 2:23 PM, Ikai L (Google) <[email protected]>wrote:
>
> >> I finished running the index tool. This is very strange. I'm seeing
> >> exactly what you're seeing: the 2011 entity is the first one returned. Let
> >> me chase this down for you.
>
> >> On Wed, Jul 14, 2010 at 12:24 PM, Ikai L (Google) <[email protected]>wrote:
>
> >>> Yeah, I'm checking for index corruption. It could also mean there's a bug
> >>> in the way we're sorting.
>
> >>> On Wed, Jul 14, 2010 at 11:51 AM, Daniel 
> >>> <[email protected]>wrote:
>
> >>>> Thanks for looking.  If you run the query in the admin console:
> >>>> "SELECT * FROM Keyword WHERE is_active = True ORDER BY last_check ASC"
> >>>> you will see the entity with key
> >>>> agtsb294aWktYmV0YXIaCxIHS2V5d29yZCINa2V5X3N5bnRoZXNpbww appear as the
> >>>> first record although the date is set to 2011.  This indicates that
> >>>> the entity is not returning as it should.  This has nothing to do with
> >>>> my code.
>
> >>>> I don't understand how your admin console is returning something
> >>>> different from mine.
>
> >>>> On Jul 14, 2:22 pm, "Ikai L (Google)" <[email protected]> wrote:
> >>>> > Daniel, I'm going to try running a tool on your data. I'll email this
> >>>> group
> >>>> > again when it's finished. I'm curious as to whether or not this does
> >>>> > anything for you. My testing via the admin console (I don't have
> >>>> access to
> >>>> > your code) seems to indicate that everything is and was working
> >>>> correctly,
> >>>> > but I'd like to establish a greater confidence level either way.
>
> >>>> > On Wed, Jul 14, 2010 at 8:03 AM, Daniel <[email protected]>
> >>>> wrote:
> >>>> > > I would like a reply...
>
> >>>> > > I was able to delete and re-add this record and that fixed that
> >>>> query,
> >>>> > > but it turns out I have another record that is not updating and
> >>>> can't
> >>>> > > be deleted.
>
> >>>> > > This record key is
> >>>> > > agtsb294aWktYmV0YXIaCxIHS2V5d29yZCINa2V5X3N5bnRoZXNpbww the kind is
> >>>> > > Keyword.
>
> >>>> > > It appears that the record broke during the same downtime last week.
> >>>> > > It always show the current time and cannot be changed in the admin
> >>>> > > console or by code.  I cannot delete this entity.
>
> >>>> > > It seems my data is going corrupt all over the place...
>
> >>>> > > On Jul 10, 10:59 am, Daniel <[email protected]> wrote:
> >>>> > > > Okay, regardless of your ability to offer me any sort of instant
> >>>> > > > support, I need this problem fixed.
>
> >>>> > > > I am not setting any read consistency. The query that you
> >>>> specified
> >>>> > > > (SELECT * FROM Search_update ORDER BY last_check DESC LIMIT 10) is
> >>>> set
> >>>> > > > to DESC, so of course it appears correct that the 2011 record is
> >>>> > > > first, but when you change it to ASC the 2011 record still appears
> >>>> > > > first.  The broken entity
> >>>> > > > 'agtsb294aWktYmV0YXIVCxINU2VhcmNoX3VwZGF0ZRi62AUM' now has changed
> >>>> to
> >>>> > > > 2011 from what you did yesterday but is is showing up as the
> >>>> OLDEST
> >>>> > > > record.  if you look at the list you will see that after the
> >>>> broken
> >>>> > > > record, the following 9 records are in ascending order and set to
> >>>> > > > 2010.  My original query above specified ascending order.
>
> >>>> > > > agtsb294aWktYmV0YXIVCxINU2VhcmNoX3VwZGF0ZRi62AUM 2011-07-09
> >>>> > > > 17:57:37.894714
> >>>> > > > agtsb294aWktYmV0YXIVCxINU2VhcmNoX3VwZGF0ZRiUvi8M 2010-07-10
> >>>> > > > 12:14:52.306180
> >>>> > > > agtsb294aWktYmV0YXIVCxINU2VhcmNoX3VwZGF0ZRiawS4M 2010-07-10
> >>>> > > > 12:15:18.649276
> >>>> > > > agtsb294aWktYmV0YXIWCxINU2VhcmNoX3VwZGF0ZRjSkLEBDA 2010-07-10
> >>>> > > > 12:15:40.274290
> >>>> > > > agtsb294aWktYmV0YXIVCxINU2VhcmNoX3VwZGF0ZRiw_gUM 2010-07-10
> >>>> > > > 12:15:43.913061
> >>>> > > > agtsb294aWktYmV0YXIVCxINU2VhcmNoX3VwZGF0ZRjw1gUM 2010-07-10
> >>>> > > > 12:15:52.877386
> >>>> > > > agtsb294aWktYmV0YXIVCxINU2VhcmNoX3VwZGF0ZRif7FkM 2010-07-10
> >>>> > > > 12:15:54.880721
> >>>> > > > agtsb294aWktYmV0YXIVCxINU2VhcmNoX3VwZGF0ZRipqgUM 2010-07-10
> >>>> > > > 12:15:56.204075
> >>>> > > > agtsb294aWktYmV0YXIVCxINU2VhcmNoX3VwZGF0ZRiv4AUM 2010-07-10
> >>>> > > > 12:15:59.092526
> >>>> > > > agtsb294aWktYmV0YXIVCxINU2VhcmNoX3VwZGF0ZRijz20M 2010-07-10
> >>>> > > > 12:16:00.490951
>
> >>>> > > > In general I like google app engine, but I am getting frustrated
> >>>> with
> >>>> > > > it's ability to consistently perform.  You have to realize that I
> >>>> have
> >>>> > > > an app that is coded correctly and was running fine until during
> >>>> the
> >>>> > > > planned downtime three days ago when this query stopped correctly
> >>>> > > > returning data, this is a structural flaw and completely disrupts
> >>>> my
> >>>> > > > ability to provide a service on your system. Also, this is the
> >>>> second
> >>>> > > > time this has happened and I had to go through quite a bit to fix
> >>>> it
> >>>> > > > the first time.  It seems that I have to repeat my last efforts,
> >>>> and
> >>>> > > > with no hope that it will not happen again and require me to
> >>>> generate
> >>>> > > > 2 days more worth of posts to get you to fix the problem.
>
> >>>> > > > On Jul 9, 5:48 pm, Ikai Lan <[email protected]> wrote:
>
> >>>> > > > > I manually reverted it to 2010 because I didn't want to cause
> >>>> problems
> >>>> > > with
> >>>> > > > > your app by fudging with the data. It didn't revert
> >>>> automatically. I
> >>>> > > just
> >>>> > > > > ran this GQL query:
>
> >>>> > > > > SELECT * FROM Search_update ORDER BY last_check DESC LIMIT 10
>
> >>>> > > > > And it displayed the data correctly.
>
> >>>> > > > > Is there any chance you're setting a different read consistency?
>
> >>>>http://code.google.com/appengine/docs/python/datastore/queriesandinde.
> >>>> > > ..
>
> >>>> > > > > I understand your frustration with regards to paid support - but
> >>>> our
> >>>> > > pricing
> >>>> > > > > structure doesn't bundle in a guarantee for a support response
> >>>> time. We
> >>>> > > make
> >>>> > > > > every effort to address any issues in production either behind
> >>>> the
> >>>> > > scenes,
> >>>> > > > > through proactive monitoring or to reactive reports via the
> >>>> different
> >>>> > > > > channels we keep track of, but we simply can't scale up a direct
> >>>> > > support
> >>>> > > > > option for developers who want to be able to pick up the red
> >>>> phone and
> >>>> > > get
> >>>> > > > > someone on the line immediately when there are service issues.
> >>>> In terms
> >>>> > > of
> >>>> > > > > addressing service degradation without additional cost to
> >>>> developers,
> >>>> > > based
> >>>> > > > > on conversations we have had with developers, I believe that we
> >>>> are as
> >>>> > > good
> >>>> > > > > if not much better than the industry standard for platform,
> >>>> managed
> >>>> > > cloud
> >>>> > > > > hosting. There are companies who will always have a resource for
> >>>> you
> >>>> > > > > available, but there's a reason any kind of baseline production
> >>>> spend
> >>>> > > will
> >>>> > > > > be in the hundreds of dollars - minimum. We're doing our best to
> >>>> > > minimize
> >>>> > > > > costs, support all developers on our platform, all while not
> >>>> charging
> >>>> > > > > developers who don't require a response time SLA.
>
> >>>> > > > > On Fri, Jul 9, 2010 at 2:24 PM, Daniel <
> >>>> [email protected]>
> >>>> > > wrote:
> >>>> > > > > > When I run the query
> >>>> > > Search_update.all().order('last_check').fetch(10)
> >>>> > > > > > programtically I consistently get this response:
>
> >>>> > > > > > agtsb294aWktYmV0YXIVCxINU2VhcmNoX3VwZGF0ZRi62AUM 2010-07-09
> >>>> > > > > > 17:57:37.894714
> >>>> > > > > > agtsb294aWktYmV0YXIVCxINU2VhcmNoX3VwZGF0ZRjsm3kM 2010-07-08
> >>>> > > > > > 00:27:52.385926
> >>>> > > > > > agtsb294aWktYmV0YXIVCxINU2VhcmNoX3VwZGF0ZRipkjsM 2010-07-08
> >>>> > > > > > 00:27:53.689982
> >>>> > > > > > agtsb294aWktYmV0YXIWCxINU2VhcmNoX3VwZGF0ZRiqu7ABDA 2010-07-08
> >>>> > > > > > 00:27:54.616698
> >>>> > > > > > agtsb294aWktYmV0YXIVCxINU2VhcmNoX3VwZGF0ZRiTrVwM 2010-07-08
> >>>> > > > > > 00:27:55.453521
> >>>> > > > > > agtsb294aWktYmV0YXIVCxINU2VhcmNoX3VwZGF0ZRjksQUM 2010-07-08
> >>>> > > > > > 00:27:56.437609
> >>>> > > > > > agtsb294aWktYmV0YXIWCxINU2VhcmNoX3VwZGF0ZRjp_KwBDA 2010-07-08
> >>>> > > > > > 00:28:02.699297
> >>>> > > > > > agtsb294aWktYmV0YXIVCxINU2VhcmNoX3VwZGF0ZRiRqAUM 2010-07-08
> >>>> > > > > > 00:28:04.598150
> >>>> > > > > > agtsb294aWktYmV0YXIWCxINU2VhcmNoX3VwZGF0ZRjyzK8BDA 2010-07-08
> >>>> > > > > > 00:28:05.595406
> >>>> > > > > > agtsb294aWktYmV0YXIVCxINU2VhcmNoX3VwZGF0ZRi-2AUM 2010-07-08
> >>>> > > > > > 00:28:07.950788
>
> >>>> > > > > > Which places the record in question first in the query
> >>>> although it
> >>>> > > has
> >>>> > > > > > a date later than the listed records.  You can also see that
> >>>> having
> >>>> > > > > > just run the response that the 2011 date reverted.  It appears
> >>>> to
> >>>> > > > > > change when you save it, but it doesn't actually change.
>
> >>>> > > > > > On Jul 9, 5:14 pm, "Ikai L (Google)" <[email protected]>
> >>>> wrote:
> >>>> > > > > > > Using this GQL query in the admin console:
>
> >>>> > > > > > > SELECT * FROM Search_update where
>
> >>>> __key__=KEY('agtsb294aWktYmV0YXIVCxINU2VhcmNoX3VwZGF0ZRi62AUM')
>
> >>>> > > > > > > I was able to set the year to 2011, see the changes
> >>>> reflected, then
> >>>> > > set
> >>>> > > > > > them
> >>>> > > > > > > back to...
>
> read more »

-- 
You received this message because you are subscribed to the Google Groups 
"Google App Engine" group.
To post to this group, send email to [email protected].
To unsubscribe from this group, send email to 
[email protected].
For more options, visit this group at 
http://groups.google.com/group/google-appengine?hl=en.

Reply via email to