Yes, kaha persistance stores directly to the disk, but last time I've checked it (few days ago) it was still using memory somehow until OutOfMemoryException, also I'm not sure about kaha maturity.
I was thinking maybe it's possible to implement that if some producer hits memory limit on journal store, it can trigger checkpoint instead of wiating for scheduled checkpoint. James.Strachan wrote: > > If you are using the journal with JDBC then the broker needs to keep > around the messages in RAM until the checkpoint (as it needs to write > them to the database); though AFAIK if you are using kaha then the > journal is the persistent store - so AFAIK the broker can evict those > messages from RAM without a checkpoint > > On 1/12/07, Danielius Jurna <[EMAIL PROTECTED]> wrote: >> >> It seems, that message spooling for persistent queue messages is already >> implemented long time ago (in 4.0.x). But messages are spooled only >> during >> checkpoint (which happens every 5 minutes or so). >> Is it possible to configure how often messages are checkpointed? >> Between checkpoints producers are blocked, if queue is full. It seems >> that >> there was some kind of fix, witch doesn't block the producer, but spools >> messages directly to disk. But it was only for non-persistent messages >> right? Is it possible to implement the same in journaled store. If queue >> is >> full, instead of blocking producer, put messages directly to persistent >> store, or trigger checkpoint. >> I was testing using 4.2 latest snapshot. >> >> -- >> View this message in context: >> http://www.nabble.com/Message-spooling-on-persistent-queues-tf2964712.html#a8294775 >> Sent from the ActiveMQ - User mailing list archive at Nabble.com. >> >> > > > -- > > James > ------- > http://radio.weblogs.com/0112098/ > -- View this message in context: http://www.nabble.com/Message-spooling-on-persistent-queues-tf2964712.html#a8295212 Sent from the ActiveMQ - User mailing list archive at Nabble.com.
