AFAIK ActiveMQ writes the persistent message to disk prior to acknowledging receipt to the producer. That's relatively simple.
What do you mean by "system failure"? You mean the s/w dies or something more serious like filesystem corruption? Not entirely certain what your question really is. On 13 September 2010 18:11, nervousbadger <nervousbad...@googlemail.com> wrote: > > Hi, > > We are currently in the process of migrating to ActiveMQ from our previous > JMS provider. We hit a big performance issue with message production speeds > on a blade server (we didn't see this during initial testing on a non-blade > server, we suspect it is due to the disk caching in use) - forcing async > send gets round the performance issue, but we are trying to work out if > there is any way of calculating the maximum number of messages that could be > lost on system failure. > > We know the size of the messages, is there any way of getting from this to a > maximum number of messages that could be lost? > > We can't move to transacted sessions just yet, as this will require code > changes across many components of our system. > > Thanks, > Pete. > -- > View this message in context: > http://activemq.2283324.n4.nabble.com/ActiveMQ-sending-async-persistent-messages-what-is-the-maximum-that-can-be-lost-tp2537793p2537793.html > Sent from the ActiveMQ - User mailing list archive at Nabble.com. >
