[jbosscache-dev] AsyncCacheLoader state transfer problems

Manik Surtani manik at jboss.org
Thu Aug 31 12:06:05 EDT 2006


Yes, I suspect state transfer methods may have to be done in the same  
thread otherwise we will have issues.
--
Manik Surtani

Lead, JBoss Cache
JBoss, a division of Red Hat

Email: manik at jboss.org
Telephone: +44 7786 702 706
MSN: manik at surtani.org
Yahoo/AIM/Skype: maniksurtani


On 31 Aug 2006, at 16:36, Vladimir Blagojevic wrote:

> Hi,
>
> Working closely with cachloaders in the last two weeks I've noticed  
> that
> current implementation of AsyncCacheLoader, if I understand things
> correctly, might lead to potenntial problems.
>
> If you recall, jbc state transfer (both streaming and byte based)
> algorithm is basically:
>
> -acquire lock on root
> -transfer state (receive or send)
> -release lock on root
>
> Transfer state part of this algorithm is done through cacheloader API
> calls (loadState/storeState etc). Besides lock
> acquisition/transfer/release there is also a lot of stream creation,
> cleanup, etc etc. This algorithm is good as long as we are running on
> one thread.
>
>
> AsyncCacheLoader does not conform to these assumptions since it  
> runs all
> operations on its own thread. It seems like that we have to exempt
> load/store state operations from being done on separate, asynch
> cacheloader thread.
>
> Any thoughts?
>
>
> _______________________________________________
> jbosscache-dev mailing list
> jbosscache-dev at lists.jboss.org
> https://lists.jboss.org/mailman/listinfo/jbosscache-dev




More information about the jbosscache-dev mailing list