CAS5, Hazelcast clustering question?

[Yan Zhou]

Hello, 

CAS 5.3.x, using Hazelcast for ticket registry.  Two CAS instances on the same server A. 

I was trying to avoid port conflict in cas.properties, so,  one instances' cas.properties look like this 

cas.ticket.registry.hazelcast.cluster.members=A

cas.ticket.registry.hazelcast.cluster.instanceName=A

cas.ticket.registry.hazelcast.cluster.port=5701

the other one look like this (note the port# difference):

cas.ticket.registry.hazelcast.cluster.members=A

cas.ticket.registry.hazelcast.cluster.instanceName=A

cas.ticket.registry.hazelcast.cluster.port=5703

Would these two hazelcast form a cluster?  My experiment says they do not, as shown in "Members" output seen in the log.

Obviously, both hazelcast cannot bind on port 5701 at the same time, the right way to specify cluster members is to use SAME port and let Hazelcast auto-increment feature to take place?

In other words, cas.properties on both instances should read like this:

cas.ticket.registry.hazelcast.cluster.members=A

cas.ticket.registry.hazelcast.cluster.instanceName=A

#

## or simply comment this line out

#

cas.ticket.registry.hazelcast.cluster.port=5701  

Thx!

Yan

[Ray Bon]

Yan,

Do you mean they both try to come up with 5701 or the second one complains the port is already taken?

Perhaps hazelcast tries to grab selection of ports. Maybe set one to 4701.

Ray

-- 

Ray Bon

Programmer Analyst

Development Services, University Systems

2507218831 | CLE 019 | rb...@uvic.ca

I respectfully acknowledge that my place of work is located within the ancestral, traditional and unceded territory of the Songhees, Esquimalt and WSÁNEĆ Nations.

[Yan Zhou]

Hi, 

I am saying, in order for Hazelcast instance to auto-join cluster, it seems that the property file must specify the same port#.  So, first instance running takes 5701, and 2nd instance tries to bind 5701, it cannot, so it goes to 5702, this process tells that the two instances are two members of the same cluster.  

If I specify two different ports 5701 and 5702 for two instances, to begin with, the two instances are still binding to 5701 and 5702 respectively, but they are not in the same cluster, there would be two hazelcast clusters (each with only one member in it). I see that in "Members" log output. 

Do I have that right?  That seems to be our problem.  Because we have ticket from one instance that cannot be validated in the other instance.  My correction is Not to specify different port# in cas.properties.

Thx!