On Wed, Aug 6, 2008 at 5:33 PM, Jared Graber <[EMAIL PROTECTED]> wrote: > I would, but I need it to run in a transaction so I can't avoid the transient > space so easily.
why? the transient space has nothing to do with the transaction support. cheers stefan > > -Jared > > -----Original Message----- > From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Stefan > Guggisberg > Sent: Wednesday, August 06, 2008 10:50 AM > To: [email protected] > Subject: Re: InMemPersistenceManager hogging resources > > On Wed, Aug 6, 2008 at 4:15 PM, Jared Graber <[EMAIL PROTECTED]> wrote: >> As it turns out, a bunch of the persistent managers use a boatload of memory >> during an import. >> The InMemPersistenceManager is probably the worst (as expected), but even >> the Oracle9PersistenceManager (a bundle persistence manager) uses a few >> hundred MB. > > i don't think this is a PM-related issue. it is true that the > Bundle*PMs use an internal cache. however, > the default size of this 'bundle cache' is just 8mb... i rather assume > the heap is used by the > transient space/change log during the import. make sure you're using > the Workspace.importXML method > rather than Session.importXML (avoiding the transient space). the > change log is still memory bound. > > cheers > stefan > >> >> -Jared >> >> -----Original Message----- >> From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Stefan >> Guggisberg >> Sent: Wednesday, August 06, 2008 5:56 AM >> To: [email protected] >> Subject: Re: InMemPersistenceManager hogging resources >> >> On Mon, Aug 4, 2008 at 9:18 PM, Jared Graber <[EMAIL PROTECTED]> wrote: >>> It seems to me that this renders the InMemPersistenceManager unusable >>> except for unit testing (which is works really well for). >> >> http://jackrabbit.apache.org/api/1.3/org/apache/jackrabbit/core/persistence/mem/InMemPersistenceManager.html >> >> please note the last sentence (bold) of the class description ;) >> >> cheers >> stefan >> >>> None of the properties contain a lot of data - mostly 10 chars or less, in >>> rare cases (less than 100 nodes) they would break the 50 character mark and >>> even then it's probably maxing out at around 500 characters. I'm not sure >>> data compression would really help. >>> >>> It looks like there is a great deal of memory overhead with the jackrabbit >>> objects because there isn't that much actual data. >>> >>> -Jared >>> >>> -----Original Message----- >>> From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Thomas Müller >>> Sent: Monday, August 04, 2008 2:50 PM >>> To: [email protected] >>> Subject: Re: InMemPersistenceManager hogging resources >>> >>> Hi, >>> >>>> When I exported the content to an XML file, it was about 15MB. >>>> When I imported the data into the repository my webapp climbs to 800MB >>>> Is this normal? >>> >>> I think yes, unfortunately. One idea to solve is to add data >>> compression to the InMemPersistenceManager. I'm not sure how much that >>> would save however. >>> >>> Regards, >>> Thomas >>> >> >
