top of page

What a T24 Expert must know about Pool Sizing

A critical aspect of optimizing the performance of T24 is to properly size the connection and thread pools. When the pool size is too small, it can lead to performance issues such as slow response times and even system crashes. Conversely, setting a too-large pool size can lead to excessive memory usage and slower performance.


Too many child processes


Please remember that the number of jbase_agents spawned is determined by the min/max pool setting in the JCA connection factory's configuration properties.

If the minimum session is set to "1," only one JBase agent will be started at JBase startup. Based on the request flow to T24, it will increase to the entire session and remain the same until the system becomes idle. Jbase agent sessions can be killed after the timeout setting if parameters such as "idle-timeout-minutes" are set.

Max pool size in JBoss EAP with TOCFEE


We configure the pool size at the MDB level. The JBoss-ejb3.xml configuration incorporates the maximum pool size.tocfee.ear\tocfOFSmessage-ejb.jar\META-INF\. When reaching the maximum limit, the number of TAFC agents will be one more than those specified at the MDB level. The reason is that when we start a TAFC agent, one agent will get started no matter whether any requests come in or not. Subsequently, the number will increase as the TAFC resources start getting used.

Keep the jbase agent session running without having it shut down.


Kindly be informed that the connection pools and killing or refreshing of the connections will be controlled and maintained by the JBoss Application Server. Therefore, no involvement will be involved in supporting the connection pools by our T24 components, such as tocfT24ra-ra.rar and tocfee.ear.

Delay in processing transactions through JBOSS when the session is made active from the idle state


We understand the delay in the first transaction processed via JBoss from the issue description. Therefore, please be informed that it is evident that there will be some delay in the first transaction processing in terms of JCA connection establishment and T24 reading the values directly from T24.

"No Managed Connection Available Within the Configured Blocking Timeout of 30000 ms."


The issue will only occur in JBoss EAP 7, which uses ActiveMQ as a JMS provider by default. Therefore, the problem is not with TAFJ but JBoss 7EAP's ActiveMQ configuration.

TC Server version

Some aspects determining the TC Server's performance are:

  • TC server multithreading on a transaction level

  • The transaction rate per second.

  • Maximum transaction capacity for concurrent transactions


Please get in touch with our team anytime if you need assistance to size the database, thread, and other pools in your core banking system accordingly.


Keep up the great work! Happy performance engineering!




111 views0 comments

Comments


bottom of page