Thanks, I'll definitely run it through appstats as soon as I get a
chance...
...here's the code, though, if you'd like to take a look. Nothing too
crazy. I'm using Objectify.
Objectify ofy = new DAO().ofy();
ArrayList<Key<FakeEntity>> arr = new ArrayList<Key<FakeEntity>>();
Query<FakeEntity> query =
ofy.query(FakeEntity.class).filter("lookAtMe", "woo");
QueryResultIterator<FakeEntity> iterator = query.iterator();
while(iterator.hasNext()){
arr.add(iterator.next().getKey());
}
ofy.delete(arr);
Essentially just gets an objectify object, does a query over my test
object "FakeEntity" looking for all entries whose "lookAtMe" field is
set to "woo," iterates over them, adding to a list, and then passes
the list to the delete call.
I know iterating over each one is not necessary with Objectify and I
can simply pass the iterator to the delete call, but my test is
simulating what the real code will need to do (i.e. look at each
object before determining if it can be deleted or not.)
Nothing too crazy, but there may be some fundamental misunderstanding
of how bulk deletes should be done.
On Feb 7, 12:34 pm, Robert Kluin <[email protected]> wrote:
> Sometimes it can be more expensive than expected, but there is also a
> fair chance your code is missing some 'basic' (ie common app engine)
> optimizations. Use Appstats to profile your code, look for lots of
> RPC calls done in serial. Make sure you read about keys only queries
> and batch operations.
>
> Also, post your test code. If there is low-hanging-fruit, people will
> (probably) let you know about it pretty quickly.
>
> http://code.google.com/appengine/docs/java/tools/appstats.htmlhttp://code.google.com/appengine/docs/python/tools/appstats.html
>
> Robert
>
>
>
> On Mon, Feb 7, 2011 at 11:54, HalcyonDays <[email protected]> wrote:
> > Hello all -
>
> > I've been doing a few load tests with the task queue, and I ran across
> > some interestingly large numbers... I constructed a very simple test
> > to put batches of a thousand or so fake items into the datastore at a
> > time, and then created a task to see how many I could query, count,
> > and remove in a reasonable amount of time. Wall (real) time was
> > fantastic for both tests, roughly 18653 ms to delete 21,000 of my test
> > items... the number that had me a teensy bit worried was the CPU
> > time... 1494101 cpu_ms. So... essentially 20-some minutes. If this
> > is normal, then it's just something I'll have to be aware of for any
> > expensive tasks... but is this standard/to be expected? It seems odd,
> > even on something as massively parallel as the datastore that I would
> > be able to rack up that much CPU time in a little under 20 seconds.
>
> > Anyway, just curious.
>
> > Thanks!
>
> > --
> > 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
> > athttp://groups.google.com/group/google-appengine?hl=en.
--
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.
