Hi Chris,
> How are you identifying contention on the transaction log?
Weblogic checks the connections in its pools and issues a rollback on those connections.
So what I often see is 50 and more processes executing a ROLLBACK command and blocked by a process that is doing a rollback, i.e. waiting for each other to complete.
> Are you using the latest jConnect EBF/SP? And have applied the sp's that come with it to the ASE?
How to find out what is the latest ?
The latest I know is what comes with ASE 15.7 SP 132.
jConnect (TM) for JDBC(TM)/7.07 SP130 (Build 27074)/P/EBF23283/JDK 1.6.0/jdbcmain/OPT/Mon Jul 21 12:39:01 PDT 2014
> What is the isolation level you are opening connections on?
Default
> What is the locking mechanism on the underlying tables? How many locks are configured? What is the performance on the log devices? Are you using a separate log cache for your 'login' database and the other database? How about 'user log cache' size? Other caches? Memory and engines? Are statistics up-to-date?
The transactions are only doing a rollback. There are no tables involved.
Thanks,
Luc.
