Hello,

There are two things you must consider. JSESSION replication, and Ticket 
Registry Replication.

If you have session-affinity configured you don't need to worry about the 
first. (It also makes for very chatty application-servers if you do).
You have to make sure though that your entire cluster is able to access the 
tickets present in the ticket registry.

If a user logs in, and a TGT and ST are created, the tickets must be available 
to all CAS servers in the cluster in case the application tries to validate it 
via any of the peers. I prefer to have active-passive configuration of the 
CAS-servers, so that all login procedures can be found in a single log-file. 
(If you don't have a proper SIEM or Log-Analyser available in your 
infrastructure that is)

Cheers,

Stephan


On 07/14/15 11:29 AM, Damiano Biagioli  <[email protected]> wrote:
> 
> 
> 
> 
> 
> 
> 
> Hello everyone ,
> 
> We are using the CAS software from about one month , but until now we have 
> never used a Clustered CAS . 
> We have deployed the CAS on two nodes that have a load balancer in front of 
> them , let's call them N1 and N2.
> 
> on both nodes there are some CASified applications App1 , App2,etc ... the 
> two nodes N1 and N2 are not part of a cluster and we Use JBoss EAP 6.3.0 as a 
> server.
> That is, on N1 a, CAS webapp, App1 and App2 are deployed , and , on node N2 , 
> there's another instance of a CAS webapp,another instance of a App1 and 
> another instance 
> 
> of App2. 
> 
> However , because the hardware load balancer provides session affinity for 
> both nodes , things work for now. 
> 
> But
> now we have to CASify another webapp (App3 ) that is deployed on another 
> server (let's call it N3) and things have become a little bit complicated . 
> We came across the problem 
> 
> described in 
> http://jasig.github.io/cas/4.0.x/planning/High-Availability-Guide.html , that 
> is , "Since both requests flow through the load balancer from different 
> 
> source addresses, it is not possible to guarantee that both requests are 
> serviced by the same CAS node" (quote from that page) .
> So I think that we need to have a clustered CAS .
> 
> I've read the following links, 
> 
> http://jasig.github.io/cas/4.0.x/installation/JPA-Ticket-Registry.html
> 
> http://jasig.github.io/cas/4.0.x/planning/High-Availability-Guide.html
> 
> https://wiki.jasig.org/display/CASUM/Clustering+CAS
> 
> and i have some questions 
> 1) have i missed some important docs i should read ? For example is there a 
> guide on how to create a CAS CLuster using jboss EAP 6.3 ? 
> 
> 2) Do we need to have session replication ? is there a guide on how to do it 
> with jboss ? (official CAS guide describes how to do session replication with 
> tomcat )
> 3) I see that we need to have a shared ticket registry in order to have a 
> clustered CAS . Do we have to use jboss cache ? i think that project is now 
> in maintenance mode ... is there some shared ticket registry that can use the 
> infinispan cache ,by chance ? 
> 4) For the shared ticket registry, do you recommend to use the 
> jpaTicketRegistry ? are there any more docs on it ? 
> 
> Thanks for any help ,
> Sorry for my poor english,
> Damiano 
> 
> 
> -- You are currently subscribed to [email protected] as: 
> [email protected] unsubscribe, change settings or access archives, see 
> http://www.ja-sig.org/wiki/display/JSG/cas-user
> 
> 
> 
> 
-- 
HE Space Operations B.V. for ESA - European Space Agency


Stephan Arts
Senior UNIX Engineer

Research & Scientific Support Department
Science & Robotic Exploration Directorate (SRE)

ESTEC
Keplerlaan 1, PO Box 299
NL-2200 AG Noordwijk, The Netherlands
[email protected] | www.esa.int
 T +31 71 565 6832 | F +31 71 565 4690

-- 
You are currently subscribed to [email protected] as: 
[email protected]
To unsubscribe, change settings or access archives, see 
http://www.ja-sig.org/wiki/display/JSG/cas-user

Reply via email to