How to trigger infinispan start in clustering environment with Spring

[scratchback]

Hi Guys,

I am using infinispan7.1 on a legacy project running on tomcat for clustering(4 nodes, A, B, C, D) cache requirement. I am using existing spring 3.2  annotation with infinispan.xml and jgroups.xml to register

<cache:annotation-driven />

<infinispan:embedded-cache-manager id="cacheManager" configuration="classpath:infinispan.xml"/>

and then Using @EnableCaching at class level and @Cachable at method level.

The issue is the binding port is not open till the @Cachable method is invoked on each node and consequently each node makes its own service call instead of getting from clustering cache.  How can I trigger the infinispan cache(open the binding# and discover the other node) during spring/web application loading instead of explicitly invoking the service?

Any comment is appreciated.

[scratchback]

Very weird, it is working change without any change(At least I did not find any change).

The difference(The second line) from the log is very obviously, for example, I made a direct call to Node A and then Node B,

Yesterday's Node B log:

INFO  o.i.r.t.jgroups.JGroupsTransport.start - ISPN000078: Starting JGroups channel ISPN

INFO  o.i.r.t.jgroups.JGroupsTransport.viewAccepted - ISPN000094:Received new cluster view for channel ISPN: [B-56020|0] (1) [B-56020]

....Some log for the same service call with the same key value but get different value(Not from Cache node A)

INFO  o.i.r.t.jgroups.JGroupsTransport.viewAccepted - ISPN000093: Received new, MERGED cluster view for channel ISPN: MergeView::[B-43665|1] (2) [B-43665, A-21243], 2 subgroups: [B-43665|0] (1) [B-43665], [A-21243|0] (1) [A-21243]

Today's Node B log:

INFO  o.i.r.t.jgroups.JGroupsTransport.start - ISPN000078: Starting JGroups channel ISPN

INFO  o.i.r.t.jgroups.JGroupsTransport.viewAccepted - ISPN000094: Received new cluster view for channel ISPN: [A-31564|1] (2) [A-31564, B-41222]

Today's log showed that once the cache is triggered it found Node A immediately but yesterday's showed it did not find Node A from the beginning.

Was it a network issue in the last two days? Or some change that I did not notice?