You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@openjpa.apache.org by Alex Snaps <al...@gmail.com> on 2011/02/03 17:58:02 UTC
EntityManagerFactory clearing caches on close()?
Hi guys,
I've not investigated that much at the code level yet, but it seems
EntityManagerFactory.close() calls AbstractQueryCache.close() which in
turn calls clearInternal(). As a result a distributed query cache ends
up cleared when a node is shut down. This only seems to happen on the
query cache though. Is it meant that way? If so, does that mean a
query cache shouldn't be clustered?
Thanks,
Alex
--
Alex Snaps <al...@gmail.com>
Senior Software Engineer - Terracotta
http://twitter.com/alexsnaps
http://www.linkedin.com/in/alexsnaps
Re: EntityManagerFactory clearing caches on close()?
Posted by Alex Snaps <al...@gmail.com>.
Rick,
Thanks a lot for clearing that for me! Just wanted to be sure there
wasn't something obvious I was missing.
Thanks again, also for your quick answers!
Alex
On Thu, Feb 3, 2011 at 6:43 PM, Rick Curtis <cu...@gmail.com> wrote:
> Alex -
>
>> Isn't it up to the Cache to make decisions about distribution or not ?
> Yes and our built-in implementation (ConcurrentQueryCache) doesn't
> distribute.
>
>> I just wonder why the QueryCache is explicitly being cleared on close().
> This is due to the AbstractQuery(Data)Cache implementation(s). Close
> delegates to clear() to free up the resources used by the cache. ... which
> makes sense when running in a single JVM and not so much when running
> distributed.
>
> @from javadoc
> AbstractQueryCache.close() -> Free the resources used by this cache.
> AbstractQueryCache.clear() -> Remove all data from this cache.
>
>> As I guess we could indeed override Abstract(Query|Data)Cache.close() to
> call close(false) rather than true by default.
> I think you should implement Cache.close() and do what is right for your
> implementation. The Abstract implementations are there for guidance, not the
> end all answer.
>
> Thanks,
> Rick
>
> On Thu, Feb 3, 2011 at 11:26 AM, Alex Snaps <al...@gmail.com> wrote:
>
>> I'm indeed "using" the ehcache plugin. I'm actually working on
>> maintaining that code base.
>> But I don't really understand what you mean with "we don't distribute
>> the QueryCache". Isn't it up to the Cache to make decisions about
>> distribution or not ? I just wonder why the QueryCache is explicitly
>> being cleared on close(). I probably should dig more into this, but
>> AbstractDataCache has the same close(boolean?) implementations. So
>> that on close a cache seems to be cleared by default. Just trying to
>> understand. As I guess we could indeed override
>> Abstract(Query|Data)Cache.close() to call close(false) rather than
>> true by default.
>> Any insight is appreciated.
>> Thanks,
>> Alex
>>
>> On Thu, Feb 3, 2011 at 6:14 PM, Rick Curtis <cu...@gmail.com> wrote:
>> > Alex -
>> >
>> > Hmmm, by any chance are you using the ehcache plugin? If so, it is
>> probably
>> > an issue with the plugin implementation. AFAIK with the built in query
>> cache
>> > we don't distribute the QueryCache.
>> >
>> > Thanks,
>> > Rick
>> >
>> > On Thu, Feb 3, 2011 at 10:58 AM, Alex Snaps <al...@gmail.com>
>> wrote:
>> >
>> >> Hi guys,
>> >> I've not investigated that much at the code level yet, but it seems
>> >> EntityManagerFactory.close() calls AbstractQueryCache.close() which in
>> >> turn calls clearInternal(). As a result a distributed query cache ends
>> >> up cleared when a node is shut down. This only seems to happen on the
>> >> query cache though. Is it meant that way? If so, does that mean a
>> >> query cache shouldn't be clustered?
>> >> Thanks,
>> >> Alex
>> >>
>> >> --
>> >> Alex Snaps <al...@gmail.com>
>> >> Senior Software Engineer - Terracotta
>> >> http://twitter.com/alexsnaps
>> >> http://www.linkedin.com/in/alexsnaps
>> >>
>> >
>>
>>
>>
>> --
>> Alex Snaps <al...@gmail.com>
>> Senior Software Engineer - Terracotta
>> http://twitter.com/alexsnaps
>> http://www.linkedin.com/in/alexsnaps
>>
>
--
Alex Snaps <al...@gmail.com>
Senior Software Engineer - Terracotta
http://twitter.com/alexsnaps
http://www.linkedin.com/in/alexsnaps
Re: EntityManagerFactory clearing caches on close()?
Posted by Rick Curtis <cu...@gmail.com>.
Alex -
> Isn't it up to the Cache to make decisions about distribution or not ?
Yes and our built-in implementation (ConcurrentQueryCache) doesn't
distribute.
> I just wonder why the QueryCache is explicitly being cleared on close().
This is due to the AbstractQuery(Data)Cache implementation(s). Close
delegates to clear() to free up the resources used by the cache. ... which
makes sense when running in a single JVM and not so much when running
distributed.
@from javadoc
AbstractQueryCache.close() -> Free the resources used by this cache.
AbstractQueryCache.clear() -> Remove all data from this cache.
> As I guess we could indeed override Abstract(Query|Data)Cache.close() to
call close(false) rather than true by default.
I think you should implement Cache.close() and do what is right for your
implementation. The Abstract implementations are there for guidance, not the
end all answer.
Thanks,
Rick
On Thu, Feb 3, 2011 at 11:26 AM, Alex Snaps <al...@gmail.com> wrote:
> I'm indeed "using" the ehcache plugin. I'm actually working on
> maintaining that code base.
> But I don't really understand what you mean with "we don't distribute
> the QueryCache". Isn't it up to the Cache to make decisions about
> distribution or not ? I just wonder why the QueryCache is explicitly
> being cleared on close(). I probably should dig more into this, but
> AbstractDataCache has the same close(boolean?) implementations. So
> that on close a cache seems to be cleared by default. Just trying to
> understand. As I guess we could indeed override
> Abstract(Query|Data)Cache.close() to call close(false) rather than
> true by default.
> Any insight is appreciated.
> Thanks,
> Alex
>
> On Thu, Feb 3, 2011 at 6:14 PM, Rick Curtis <cu...@gmail.com> wrote:
> > Alex -
> >
> > Hmmm, by any chance are you using the ehcache plugin? If so, it is
> probably
> > an issue with the plugin implementation. AFAIK with the built in query
> cache
> > we don't distribute the QueryCache.
> >
> > Thanks,
> > Rick
> >
> > On Thu, Feb 3, 2011 at 10:58 AM, Alex Snaps <al...@gmail.com>
> wrote:
> >
> >> Hi guys,
> >> I've not investigated that much at the code level yet, but it seems
> >> EntityManagerFactory.close() calls AbstractQueryCache.close() which in
> >> turn calls clearInternal(). As a result a distributed query cache ends
> >> up cleared when a node is shut down. This only seems to happen on the
> >> query cache though. Is it meant that way? If so, does that mean a
> >> query cache shouldn't be clustered?
> >> Thanks,
> >> Alex
> >>
> >> --
> >> Alex Snaps <al...@gmail.com>
> >> Senior Software Engineer - Terracotta
> >> http://twitter.com/alexsnaps
> >> http://www.linkedin.com/in/alexsnaps
> >>
> >
>
>
>
> --
> Alex Snaps <al...@gmail.com>
> Senior Software Engineer - Terracotta
> http://twitter.com/alexsnaps
> http://www.linkedin.com/in/alexsnaps
>
Re: EntityManagerFactory clearing caches on close()?
Posted by Alex Snaps <al...@gmail.com>.
I'm indeed "using" the ehcache plugin. I'm actually working on
maintaining that code base.
But I don't really understand what you mean with "we don't distribute
the QueryCache". Isn't it up to the Cache to make decisions about
distribution or not ? I just wonder why the QueryCache is explicitly
being cleared on close(). I probably should dig more into this, but
AbstractDataCache has the same close(boolean?) implementations. So
that on close a cache seems to be cleared by default. Just trying to
understand. As I guess we could indeed override
Abstract(Query|Data)Cache.close() to call close(false) rather than
true by default.
Any insight is appreciated.
Thanks,
Alex
On Thu, Feb 3, 2011 at 6:14 PM, Rick Curtis <cu...@gmail.com> wrote:
> Alex -
>
> Hmmm, by any chance are you using the ehcache plugin? If so, it is probably
> an issue with the plugin implementation. AFAIK with the built in query cache
> we don't distribute the QueryCache.
>
> Thanks,
> Rick
>
> On Thu, Feb 3, 2011 at 10:58 AM, Alex Snaps <al...@gmail.com> wrote:
>
>> Hi guys,
>> I've not investigated that much at the code level yet, but it seems
>> EntityManagerFactory.close() calls AbstractQueryCache.close() which in
>> turn calls clearInternal(). As a result a distributed query cache ends
>> up cleared when a node is shut down. This only seems to happen on the
>> query cache though. Is it meant that way? If so, does that mean a
>> query cache shouldn't be clustered?
>> Thanks,
>> Alex
>>
>> --
>> Alex Snaps <al...@gmail.com>
>> Senior Software Engineer - Terracotta
>> http://twitter.com/alexsnaps
>> http://www.linkedin.com/in/alexsnaps
>>
>
--
Alex Snaps <al...@gmail.com>
Senior Software Engineer - Terracotta
http://twitter.com/alexsnaps
http://www.linkedin.com/in/alexsnaps
Re: EntityManagerFactory clearing caches on close()?
Posted by Rick Curtis <cu...@gmail.com>.
Alex -
Hmmm, by any chance are you using the ehcache plugin? If so, it is probably
an issue with the plugin implementation. AFAIK with the built in query cache
we don't distribute the QueryCache.
Thanks,
Rick
On Thu, Feb 3, 2011 at 10:58 AM, Alex Snaps <al...@gmail.com> wrote:
> Hi guys,
> I've not investigated that much at the code level yet, but it seems
> EntityManagerFactory.close() calls AbstractQueryCache.close() which in
> turn calls clearInternal(). As a result a distributed query cache ends
> up cleared when a node is shut down. This only seems to happen on the
> query cache though. Is it meant that way? If so, does that mean a
> query cache shouldn't be clustered?
> Thanks,
> Alex
>
> --
> Alex Snaps <al...@gmail.com>
> Senior Software Engineer - Terracotta
> http://twitter.com/alexsnaps
> http://www.linkedin.com/in/alexsnaps
>