We're using JBossCache, and have recently upgraded to CAS 3.2.

We have been experiencing some stability problems during our more  
vigorous load-testing scenarios, (16+ virtual users in JMeter).  While  
our single-server instance can handle this load without problem, our  
clustered configuration experiences a slow degredation over time (10+  
hours) that ends with a large number of replication errors.

We have an extra complication in our configuration, however.  We are  
using the JAASAuthenticationHandler to authenticate to a Kerberos  
KDC.  There is a known memory leak in Sun's implementation which  
Andrew Feller pointed out.  We've been able to ameliorate the memory  
leak somewhat but it is most likely a contributing factor to our  
problems.

I'm retooling a pair of servers to look specifically for memory and GC  
issues, which should hopefully shed some light on our problems.

Brian

On Apr 29, 2008, at 1:40 PM, Benn Oshrin wrote:

> Brian Donnelly <[EMAIL PROTECTED]> wrote on April 28, 2008 3:01:16
> PM -0700:
>
> ] Please add University of California, Davis to the list of deployers.
> ] We're still working on getting a stable clustering solution in  
> place,
> ] but have been running it a production single-server mode for over a
> ] year now.
>
> Out of curiosity, what backend are you using for clustering, and what
> stability problems are you seeing?
>
> -Benn-
> _______________________________________________
> Yale CAS mailing list
> [email protected]
> http://tp.its.yale.edu/mailman/listinfo/cas

_______________________________________________
Yale CAS mailing list
[email protected]
http://tp.its.yale.edu/mailman/listinfo/cas

Reply via email to