Details
-
Bug
-
Status: Closed
-
Major
-
Resolution: Incomplete
-
5.2.0
-
None
-
None
-
ActiveMQ 5.2. Linux Redhat 64bits. Java 1.6 - multi network interfaces
Description
We are on a production system for an ERDF major project.
We have about 15 servers called tpnocp"XY"v [ "XY" = 01,02,03,04,.. ]. On these servers, some Java processes are launched. Some of them have ActiveMQ embedded brokers. Generally speaking, these brokers are linked together by about 4 different networks of brokers with duplex or not duplex network connectors. Multicast discovery mechanism is used to dynamically create brokers connections. Default Multicast IP is used with different groups.
I'm currently in a situation where
- a java process is launched on tpnoc09v server, and has an embedded broker named "BEA-tpnocp09v"
- this embedded broker dynamically discovered 4 brokers that are embedded in 4 java process that are launched on the tpnocp13v server (tpnocp13v-bus IP address is used), and connected in full duplex to them.
- a fifth process is executed on the same tpnocp13v server, has an embedded broker which has the same ActiveMQ configuration as the 4 others that are connected to the "BEA-tpnocp09v" broker.
- the "BEA-tpnocp09v" embedded broker is not connected to the fifth embedded broker.
- Multicast packets relative to this fifth broker, seem to be correctly received by tpnocp09v server (tcpdump)
- All seems stable.