Hi, This is, partly a followup to a post I made back in October, about a possible timing issue, and today explained in a separate post to the list. I think this is about what to use for a ticket registry, but maybe it's about mysql configuration. Our CAS environment is a pair of servers, using JpaTicketRegistry, with the ticket store being MySQL. MySQL is running on a 2 server cluster. A Cisco CSS is fronting for the CAS cluster, and the MySQL cluster. We have a problem where sometimes get an error as per the attached screenshot. NOW, having stared at the configuration, and scratched through the MySQL binary logs, I know what problem we have. This is caused by the service ticket being written to one node of the MySQL cluster, but then the other CAS node happens to get to do the service ticket validation, and it happens to use the other DB node, the ticket isn't found, because database replication hasn't been fast enough to bring the ticket over. The MySQL cluster servers certainly aren't overloaded, but they are fairly busy servers. Does anyone else have this sort of problem, is there anything obvious we could do to tweak MySQL server config, assuming I can persuade the team who run the server, to make the changes. It's sad, if we only had a single node DB, instead of a cluster, the we wouldn't have this problem. So can we fix our issue, or should we look at alternate ticket registries. I've been reading up on them, but their doesn't seem to be a consensus on what is best, which I guess is due to peoples different environments, and requirements. If we go with any registry that requires replication, are we at risk of repeats of the above fault?  ======================================= David Clarke Java Team Information Technology Infrastructure Division of Information Building 3K The Australian National University Canberra ACT 0200 Australia T: +61 2 6125 8390 www.anu.edu.au CRICOS Provider #00120C ======================================= |
smime.p7s
Description: S/MIME cryptographic signature
