@David Ohsie: We're actually doing this in development right now, so our servers are handling very little traffic. It turns out EhCache wasn't the problem.
It wasn't the CAS Client, either. It turned out to be session replication between the two Tomcat servers. More specifically, it was multicast failing between the two cluster members that was causing the high rate of session failures. I only found this after beginning to test the content of each object in an attempt to figure out where I was missing data. After validating the application code a few times, and testing for race conditions, I decided to do a packet trace. Turns out there was no packet trace to be found between the Tomcat servers. I tried some changes to the multicast setup but still it failed, so after doing some reading, I implemented static memberships for the Tomcat cluster servers. After restarting them I finally saw cluster traffic. The frequency of the missing session data dropped dramatically, as expected. At this point I'm going to work on session retry code to handle any failures encountered that would cause proxy ticket retrieval to fail. Thanks everyone. Ken On Thu, Aug 29, 2013 at 12:45 PM, Ohsie, David <[email protected]> wrote: > Knowing very little about your setup or ehcache, it is possible that the > cache is getting overloaded and the PGT is getting ejected from the cache > before it can be retrieved?**** > > ** ** > > ** ** > > *From:* Kenneth Erard [mailto:[email protected]] > *Sent:* Wednesday, August 28, 2013 4:03 PM > *To:* [email protected] > > *Subject:* Re: [cas-user] Missing Proxy Granting Tickets in CAS Client > Cluster**** > > ** ** > > Yes, it persists with synchronous replication.**** > > ** ** > > I made sure to add some thread sleep time to the web application code > (5000ms) just to make sure the web application wasn't beating EhCache. > With DEBUG logging on I can see the EhCache activity well before the web > application attempts to retrieve a proxy ticket. So, I would expect the > proxy granting ticket to be present in memory for use at this point.**** > > ** ** > > ** ** > > On Wed, Aug 28, 2013 at 3:16 PM, Tom Poage <[email protected]> wrote:*** > * > > On Aug 28, 2013, at 11:08 AM, Kenneth Erard <[email protected]> > wrote: > ...**** > > > not fix our problems, so I reverted to 2.2.0. EhCache is running > asynchronously. Log4j 1.2.17 and SLF4J 1.6.6 are**** > > I've been looking over the proxy protocol myself. Does the problem persist > if you switch to synchronous replication? (thinking STs here.) > > Tom. > > > -- > 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**** > > ** ** > > -- > 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**** > > -- > 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 > > -- 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
