After we upgraded from Wildfly 8.2.1.Final to Wildfly 9.0.1.Final, we started to getting a lot of warnings like the following:
WARNING [org.jgroups.protocols.TCP] (INT-1,ee,dev6.example.com:server1) JGRP000012: discarded message from different cluster hq-cluster (our cluster is ee). Sender was ad3f8046-3c95-f6d4-da13-3019d931f9e4 (received 4 identical messages from ad3f8046-3c95-f6d4-da13-3019d931f9e4 in the last 64159 ms)
The messages are for various hosts and servers at hosts. The same thing was in betas and CR versions of Wildfly, on the other hand, it wasn't in version 8. We are using TCP as a transport, however according to other resources the same is for UDP.
Does someone have a solution (of course other than increasing the severity level of logs)? Thanks.
We finally found the problem and the solution. Wildfly 9 is sending the messages for the cluster nodes and for HornetQ within the same communication channel, which seems to make collisions. We solved the problem by the creation of the second stack and dividing of the traffic between them.
For TCP, the working configuration is as follows:
<stacks default="tcp">
<stack name="tcp">
<transport type="TCP" socket-binding="jgroups-tcp"/>
<protocol type="TCPPING">
<property name="initial_hosts">
node1[7600],node1[7750],node2[7600],node2[7750]
</property>
<property name="port_range">
0
</property>
</protocol>
<protocol type="MERGE2"/>
<protocol type="FD_SOCK" socket-binding="jgroups-tcp-fd"/>
<protocol type="FD"/>
<protocol type="VERIFY_SUSPECT"/>
<protocol type="pbcast.NAKACK2"/>
<protocol type="UNICAST3"/>
<protocol type="pbcast.STABLE"/>
<protocol type="pbcast.GMS"/>
<protocol type="MFC"/>
<protocol type="FRAG2"/>
<protocol type="RSVP"/>
</stack>
<stack name="tcphq">
<transport type="TCP" socket-binding="jgroups-tcp-hq"/>
<protocol type="TCPPING">
<property name="initial_hosts">
node1[7660],node1[7810],node2[7660],node2[7810]
</property>
<property name="port_range">
0
</property>
</protocol>
<protocol type="MERGE2"/>
<protocol type="FD_SOCK" socket-binding="jgroups-tcp-hq-fd"/>
<protocol type="FD"/>
<protocol type="VERIFY_SUSPECT"/>
<protocol type="pbcast.NAKACK2"/>
<protocol type="UNICAST3"/>
<protocol type="pbcast.STABLE"/>
<protocol type="pbcast.GMS"/>
<protocol type="MFC"/>
<protocol type="FRAG2"/>
<protocol type="RSVP"/>
</stack>
</stacks>
You also need to configure HornetQ (use the proper jgroups-stack, tcphq in this case):
<broadcast-groups>
<broadcast-group name="bg-group1">
<jgroups-stack>tcphq</jgroups-stack>
<jgroups-channel>hq-cluster</jgroups-channel>
<broadcast-period>5000</broadcast-period>
<connector-ref>
http-connector
</connector-ref>
</broadcast-group>
</broadcast-groups>
<discovery-groups>
<discovery-group name="dg-group1">
<jgroups-stack>tcphq</jgroups-stack>
<jgroups-channel>hq-cluster</jgroups-channel>
<refresh-timeout>10000</refresh-timeout>
</discovery-group>
</discovery-groups>
...and of course you need to add the relevant socket-binding
into socket-binding-group
:
<socket-binding name="jgroups-tcp-hq" port="7660"/>
<socket-binding name="jgroups-tcp-hq-fd" port="7670"/>
Unfortunately, I have no experience with UDP, but I think the principle will be the same.
If you love us? You can donate to us via Paypal or buy me a coffee so we can maintain and grow! Thank you!
Donate Us With