>>> The Connection uses the acceptor's thread pool as its server executor >>>> and the InVMConnector's thread pool as its client executor. >>>> Is that correct?
hmmm.. it seems you're correct.. We could perhaps use just a single pool on InVM. InVM is mostly for application servers and embedded cases.. we can change it anyway you like. It would be best to keep the API as is (and perhaps deprecate it) and add new ones as you need. This is the kind of thing that's easier to interact on a chat..look for me on IRC @ apache-activemq on IRC.. we can later add a summary here so other can follow. On Mon, Jun 26, 2017 at 11:49 AM, Jeff Mesnil <[email protected]> wrote: > On Mon, Jun 26, 2017 at 4:04 PM, Clebert Suconic > <[email protected]> wrote: >>> * some org.apache.activemq.artemis.core.remoting.impl.invm.InVMConnector >>> => what's their use? >> >> does this answer this? or you're looking for something beyond of that? > > Something a bit beyond that. > At first glance, these threads are use to write (and flush) data sent > by the InVM connection. > >>> * some ActiveMQ-remoting-threads-ActiveMQServerImpl::serverUUID=xxxxx >>> => what's their use? >> >> >> The general pool on the server's... similar code from the previous... > > Are you sure? these threads are created by the RemotingService's > thread factory (when the RemotingServiceImpl is started). > > They are passed to the AcceptorFactory's createAcceptor method. > Netty implementation does not use it. > InVM implementation does use it. > > If I look at > https://github.com/apache/activemq-artemis/blob/b50ae5a916e613ef43a86b8b995b84dc9a069a82/artemis-server/src/main/java/org/apache/activemq/artemis/core/remoting/impl/invm/InVMConnector.java#L183-L185, > I see that the > The Connection uses the acceptor's thread pool as its server executor > and the InVMConnector's thread pool as its client executor. > Is that correct? > > to sum up, it seems to me that ActiveMQ-remoting-threads are threads > used by in-vm acceptor and > org.apache.activemq.artemis.core.remoting.impl.invm.InVMConnector by > in-vm connector. i.e. they are thread pools for resp. in-vm server and > client execution. > >> I thought those method facilities.. serviceRegistry and the general >> pool were being used by the integration at Wildfly. > > I have a long-standing task to do it (and some work in a branch about > that) but it's not in WildFly master branch yet. > > thanks for the info clebert, if you can confirm my explanation above, > I'll have all the info I need. > > jeff > > > -- > Jeff Mesnil > [email protected] > http://jmesnil.net/weblog/ -- Clebert Suconic
