Only one CAS node is used per db (MySQL). But there are several CAS Servers 
using independent schemas on the same db server.
Currently, approx. 100.000-200.000 tickets are in the registry.

As this server uses the fairly old CAS 3.2.1 and the JPATicketRegistry was 
improved in 3.3.3, it might help to update, first.

Other servers we run CAS 3.4.3.1 on do not see this problem, but they by far do 
not hold that much tickets.

For the 4.0 redesign, is there a Jira ticket existing? (CAS-782 is set fixed).
On redesign of the ticket registry, I strongly recommend to not iterate over 
all tickets but to use a finder to get expired tickets from the db directly.
This requires a redesign/addition of Ticket attributes but I think thats a way 
to avoid bottlenecks. Additionally, we do not need to deserialize all tickets.

Robert



Am 26.05.2011 um 03:49 schrieb Marvin Addison:

>> It seems the problem happens when the cleaner is running.
> 
> Do you have multiple nodes attempting to clean the registry
> simultaneously?  That's often another cause of deadlocks, but cleaning
> in general can increase deadlock frequency.  You could consider
> disabling the cleaner altogether and simply run a scheduled job to
> clean up orphaned tickets.  Since you're using rememberMe, it may be
> tricky to identify a cutoff date beyond which you're _certain_ tickets
> have expired, but I'd wager it's possible.  I'd recommend this option
> only as a last resort.
> 
> M
> 
> -- 
> 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

Reply via email to