Hi, I would love to upgrade from 0.9.6 which is what I run but InfoGlue which is the platform I'm developing uses Pluto portal driver and that is not compatible with 0.9.9 at least when I tried. The XML-parts of castor that is which it uses. Perhaps in due time or maybe you have heard of someone succeeding?
We uses cache eviction in a clustered env where the deliver instances of InfoGlue which delivers pages at tremendous speed gets update calls from the cms-part whenever something is published. So there are two or more apps updating the data. Regards Mattias Bogeblad -----Ursprungligt meddelande----- Från: Ralf Joachim [mailto:[EMAIL PROTECTED] Skickat: den 15 november 2005 18:55 Till: [email protected] Ämne: Re: [castor-user] Readonly-Transactions and locks Hi Mattias, with release 0.9.9.1 we releasex some hours ago we fixed a issue that caused read locks not to be released if a LockNotGrantedException had been thrown. Can you please check if this could be the reason for your problems. Have a look at: http://jira.codehaus.org/browse/CASTOR-1247 It would be interesting to know which release of castor you have the problem with? We also added org.castor.jdo.util.JDOUtils.closeDatabase(db) method for save closing of databases but I'm not sure if this could help with your problem. Maybe a look at the source of the method could give you some hints to improve your code. As far as I know there is no way to prevent cache eviction and read to occur concurrently. Therefore you will have to take care on that yourself. Why do you evict your cache? Is your database modified by another application or do you need some special cacheing algorithms? Regards Ralf Bogeblad Mattias schrieb: > Hi all, > > > > A simple question. I have an app where everything is readonly operations > against castor. I load and query everything in Database.READ_ONLY-mode. > > My question is - sometimes during extreme load I still get "...doesn't > exist or no lock exceptions" on a lot of objects I have read. I have > tried both with db.rollback() and db.commit before close but both seems > to want to lock the objects for some reason. > > > > Is it ok to skip both rollbacks and commits in such a situation and only > do db.close on the db? Would that help me? Is there any other tips on > the issue on read locks. This mostly happens when I do a cache-eviction > and a read simultaneously (uuhh, is that spelling right). In there a way > to stop cache eviction and read to occur concurrently? > > > > Regards > > Mattias Bogeblad > > > > > > *Mattias Bogeblad** > * > > * > ------------------------------------------------------------------------ > * > > [EMAIL PROTECTED] <mailto:[EMAIL PROTECTED]> > www.modul1.se <http://www.modul1.se> > Mobile +46 (0)702 80 05 05 > Phone +46 (0)31 748 84 00, Fax +46 (0)31 748 84 01 > Vallgatan 27, SE-411 16 Göteborg, Sweden *** > > > -- Syscon Ingenieurbüro für Meß- und Datentechnik GmbH Ralf Joachim Raiffeisenstraße 11 D-72127 Kusterdingen Germany Tel. +49 7071 3690 52 Mobil: +49 173 9630135 Fax +49 7071 3690 98 Email: [EMAIL PROTECTED] Web: www.syscon-world.de ------------------------------------------------- If you wish to unsubscribe from this list, please send an empty message to the following address: [EMAIL PROTECTED] ------------------------------------------------- ------------------------------------------------- If you wish to unsubscribe from this list, please send an empty message to the following address: [EMAIL PROTECTED] -------------------------------------------------
