Luiz Cordeiro wrote:
> Thanks for the answers, they were precise and fast.
>
> We're considering using QPid in a corporate project but there're a couple of points I
would like to make clearer:
>
> - Is there a way to specify a list of brokers for the C++ client lib and have the traffic
automatically distributed among them? Such that, should one of them go down, the client lib
would automatically redistribute the load for the remaining brokers? In other words, is the
C++ client lib capable of load balancing and unavailability detection? I've read about the
possibility to set a keepalive on the socket, and that the Connection would be cut in the
occurrence of 2 failures in the heartbeat process, but that was all that I could find about
it on the project's Docs.
>
The Qpid broker (c++) supports clustering whereby multiple broker nodes
are tied together to provide high availability. When using clustering,
a client connects to one of the nodes and will seamlessly fail over to
another if that node fails.
Qpid clustering is targeted at high availability and not load
balancing. The only resource that scales up is connection sockets. If
you are going to try the clustering capability, I recommend that you use
the code from the Subversion trunk. Clustering has been improved
significantly since the last release (0.5).
> - Are there SNMP consumers for QPid's event queue, such that we could monitor QPid by
integrating it with our Nagios infrastructure?
>
I'm not sure what you mean by "event queue", but log events can be
directed to Syslog by using the --log-to-syslog command line option. Or
are you asking if the broker can be monitored via SNMP (it cannot
currently).
> Thanks,
> Acácio
-Ted
---------------------------------------------------------------------
Apache Qpid - AMQP Messaging Implementation
Project: http://qpid.apache.org
Use/Interact: mailto:users-subscribe@qpid.apache.org
|