Clustering nodes between two server have high traffic CAS 5.2.0-RC4

[Andy Ng]

Hi all,

Background:

Today I have deployed our CAS 5 server (5.2.0-RC4) to production, to replace the existing CAS 4, we are using a load balancer with 2 clustering node by Hazelcast.

Problem:

there is a problem of increasing network traffic (a near straight slope) since the next morning of the deployment. It has been 8 hours since the increase of traffic and the traffic haven't slow downed.

What I tried:
I have tried shut down one node, and the traffic on the other node immediately drop down to near nothing.

However, when I resume that node, so 2 nodes running again, the traffic jump straight back to very high.

What I thinked may happens:

1. Too much tickets and they passed between the two nodes using hazelcast



2. Possible logs passing between two nodes.
Our system maintainer have told me that seem like the two clustered node is exchanging logs between each other. Below are some of the extracted logs:
____________________________________________________________________________________________________

....xsr.Borg.apereo.cas.ticket.support.TicketGrantingTicketExpirationPolicyjs8........J..maxTimeToLiveInSecond
 successfulAuthenticationHandlerssr..java.util.LinkedHashSe
..............................................................sr.'org.apereo.cas.ticket.ServiceTicketImpl.c.........Z..fromNewLoginL..grantedTicketAlready
____________________________________________________________________________________________________

Point 2 seems weird to me, why they need to communicate the logs.




Did someone else have similar experience / knows what happens / knows how to fix this? thanks!

-Andy

#My properties:

#Assume 123.0.0.1 is my first node, 123.0.0.2 is my seconds

cas.ticket.registry.hazelcast.cluster.instanceName=123.0.0.1

cas.ticket.registry.hazelcast.cluster.members=123.0.0.1,123.0.0.2

cas.ticket.registry.hazelcast.crypto.signing.key=ASD

cas.ticket.registry.hazelcast.crypto.encryption.key=ASDFFF

#No other Setting should affect on this problem, at least with my understanding

[Ray Bon]

Andy,

Have you configured a caching mechanism for the logs?

Do you have your application servers (tomcat) clustered?

CAS 5.x has good session management and does not need to be on a clustered environment.

Ray

-- 
Ray Bon
Programmer analyst
Development Services, University Systems
2507218831 | CLE 019 | rb...@uvic.ca

[Andy Ng]

Hi Rbon,

Thanks for the reply! Since the traffic has dropped down now, so I think it might be a normal thing, just real people using CAS I guess will have this behavior.

I don't think I do any cache for logs.
Tomcat is not clustered.

Thanks again!

-Andy