You are viewing a plain text version of this content. The canonical link for it is here.
Posted to ojb-dev@db.apache.org by Joseph Elliott <jf...@pobox.com> on 2003/06/24 04:51:24 UTC

deadlocking issues with LoadedObjectsRegistry use of WeakHashMap

has anybody else encountered deadlocking issues under load with the ojb
library?  we started seeing frequent situations with pegged cpu usage and
stuck threads testing a large web application (running under tomcat on duel
cpu linux boxes)  a thread dump showed numerous threads all stuck in the
WeakHashMap.put's called from LoadedObjectsRegistry.java

adding synchronize blocks around the LoadedObjectsRegistry.java put and
clear calls has seemed to have stabilized things.  does any body have any
opinions about the performance impact of this and if there are any other
alternatives to making the system safe under highly concurrent usage?

we're currently working with a patched version of ojb for our application,
but it would be nice if this could be addressed in the official code base
for the next rc.

thanks much,
 - joe


Re: deadlocking issues with LoadedObjectsRegistry use of WeakHashMap

Posted by Thomas Mahler <th...@web.de>.
Hi Joseph,

To include your changes into the codebase please post your patch (or 
patched files).
YOu can post it as a zipped file either to the list or directly to me.
Of course we want to get OJB free of such issues!

cheers,
Thomas

Joseph Elliott wrote:
> has anybody else encountered deadlocking issues under load with the ojb
> library?  we started seeing frequent situations with pegged cpu usage and
> stuck threads testing a large web application (running under tomcat on duel
> cpu linux boxes)  a thread dump showed numerous threads all stuck in the
> WeakHashMap.put's called from LoadedObjectsRegistry.java
> 
> adding synchronize blocks around the LoadedObjectsRegistry.java put and
> clear calls has seemed to have stabilized things.  does any body have any
> opinions about the performance impact of this and if there are any other
> alternatives to making the system safe under highly concurrent usage?
> 
> we're currently working with a patched version of ojb for our application,
> but it would be nice if this could be addressed in the official code base
> for the next rc.
> 
> thanks much,
>  - joe
> 
> 
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: ojb-dev-unsubscribe@db.apache.org
> For additional commands, e-mail: ojb-dev-help@db.apache.org
> 
>